Text Generation
Transformers
Safetensors
qwen3
llama-factory
full
Generated from Trainer
conversational
text-generation-inference
Instructions to use laion/r2egym-stackseq with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use laion/r2egym-stackseq with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="laion/r2egym-stackseq") messages = [ {"role": "user", "content": "Who are you?"}, ] pipe(messages)# Load model directly from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("laion/r2egym-stackseq") model = AutoModelForCausalLM.from_pretrained("laion/r2egym-stackseq", device_map="auto") messages = [ {"role": "user", "content": "Who are you?"}, ] inputs = tokenizer.apply_chat_template( messages, add_generation_prompt=True, tokenize=True, return_dict=True, return_tensors="pt", ).to(model.device) outputs = model.generate(**inputs, max_new_tokens=40) print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:])) - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use laion/r2egym-stackseq with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "laion/r2egym-stackseq" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "laion/r2egym-stackseq", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/laion/r2egym-stackseq
- SGLang
How to use laion/r2egym-stackseq with SGLang:
Install from pip and serve model
# Install SGLang from pip: pip install sglang # Start the SGLang server: python3 -m sglang.launch_server \ --model-path "laion/r2egym-stackseq" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "laion/r2egym-stackseq", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker images
docker run --gpus all \ --shm-size 32g \ -p 30000:30000 \ -v ~/.cache/huggingface:/root/.cache/huggingface \ --env "HF_TOKEN=<secret>" \ --ipc=host \ lmsysorg/sglang:latest \ python3 -m sglang.launch_server \ --model-path "laion/r2egym-stackseq" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "laion/r2egym-stackseq", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }' - Docker Model Runner
How to use laion/r2egym-stackseq with Docker Model Runner:
docker model run hf.co/laion/r2egym-stackseq
| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 7.0, | |
| "eval_steps": 500, | |
| "global_step": 6048, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.005790387955993051, | |
| "grad_norm": 16.655819645529608, | |
| "learning_rate": 2.6446280991735537e-07, | |
| "loss": 0.6604, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3099028170108795, | |
| "step": 5, | |
| "valid_targets_mean": 7389.0, | |
| "valid_targets_min": 5627 | |
| }, | |
| { | |
| "epoch": 0.011580775911986103, | |
| "grad_norm": 16.101305822612215, | |
| "learning_rate": 5.950413223140496e-07, | |
| "loss": 0.6991, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2357616424560547, | |
| "step": 10, | |
| "valid_targets_mean": 1294.8, | |
| "valid_targets_min": 422 | |
| }, | |
| { | |
| "epoch": 0.017371163867979156, | |
| "grad_norm": 14.103935692499862, | |
| "learning_rate": 9.256198347107438e-07, | |
| "loss": 0.6739, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3202242851257324, | |
| "step": 15, | |
| "valid_targets_mean": 8539.9, | |
| "valid_targets_min": 6761 | |
| }, | |
| { | |
| "epoch": 0.023161551823972205, | |
| "grad_norm": 11.207750997832555, | |
| "learning_rate": 1.2561983471074383e-06, | |
| "loss": 0.5993, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2791058421134949, | |
| "step": 20, | |
| "valid_targets_mean": 7579.0, | |
| "valid_targets_min": 6174 | |
| }, | |
| { | |
| "epoch": 0.02895193977996526, | |
| "grad_norm": 7.945461390881987, | |
| "learning_rate": 1.5867768595041324e-06, | |
| "loss": 0.5893, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.279413640499115, | |
| "step": 25, | |
| "valid_targets_mean": 7328.2, | |
| "valid_targets_min": 4839 | |
| }, | |
| { | |
| "epoch": 0.03474232773595831, | |
| "grad_norm": 5.946728043074954, | |
| "learning_rate": 1.917355371900827e-06, | |
| "loss": 0.5752, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3025493621826172, | |
| "step": 30, | |
| "valid_targets_mean": 6838.2, | |
| "valid_targets_min": 5034 | |
| }, | |
| { | |
| "epoch": 0.04053271569195136, | |
| "grad_norm": 4.920950696141524, | |
| "learning_rate": 2.247933884297521e-06, | |
| "loss": 0.552, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2719684839248657, | |
| "step": 35, | |
| "valid_targets_mean": 7248.2, | |
| "valid_targets_min": 5567 | |
| }, | |
| { | |
| "epoch": 0.04632310364794441, | |
| "grad_norm": 5.087298480449038, | |
| "learning_rate": 2.578512396694215e-06, | |
| "loss": 0.4962, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23095563054084778, | |
| "step": 40, | |
| "valid_targets_mean": 6234.0, | |
| "valid_targets_min": 5603 | |
| }, | |
| { | |
| "epoch": 0.05211349160393746, | |
| "grad_norm": 3.81594656077192, | |
| "learning_rate": 2.9090909090909093e-06, | |
| "loss": 0.4756, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10880223661661148, | |
| "step": 45, | |
| "valid_targets_mean": 602.0, | |
| "valid_targets_min": 152 | |
| }, | |
| { | |
| "epoch": 0.05790387955993052, | |
| "grad_norm": 1.459292133331144, | |
| "learning_rate": 3.2396694214876034e-06, | |
| "loss": 0.4222, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20699799060821533, | |
| "step": 50, | |
| "valid_targets_mean": 7299.9, | |
| "valid_targets_min": 6158 | |
| }, | |
| { | |
| "epoch": 0.06369426751592357, | |
| "grad_norm": 1.1003630444312296, | |
| "learning_rate": 3.5702479338842976e-06, | |
| "loss": 0.4171, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19563168287277222, | |
| "step": 55, | |
| "valid_targets_mean": 6140.0, | |
| "valid_targets_min": 4648 | |
| }, | |
| { | |
| "epoch": 0.06948465547191662, | |
| "grad_norm": 0.9275014204530865, | |
| "learning_rate": 3.900826446280992e-06, | |
| "loss": 0.4014, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19986401498317719, | |
| "step": 60, | |
| "valid_targets_mean": 6603.0, | |
| "valid_targets_min": 4882 | |
| }, | |
| { | |
| "epoch": 0.07527504342790967, | |
| "grad_norm": 0.7033608549870087, | |
| "learning_rate": 4.231404958677686e-06, | |
| "loss": 0.3812, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19471821188926697, | |
| "step": 65, | |
| "valid_targets_mean": 8015.0, | |
| "valid_targets_min": 6344 | |
| }, | |
| { | |
| "epoch": 0.08106543138390272, | |
| "grad_norm": 0.6854719385778002, | |
| "learning_rate": 4.56198347107438e-06, | |
| "loss": 0.3691, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17885202169418335, | |
| "step": 70, | |
| "valid_targets_mean": 6305.2, | |
| "valid_targets_min": 4622 | |
| }, | |
| { | |
| "epoch": 0.08685581933989578, | |
| "grad_norm": 0.617162174526013, | |
| "learning_rate": 4.892561983471075e-06, | |
| "loss": 0.3722, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18935421109199524, | |
| "step": 75, | |
| "valid_targets_mean": 7890.8, | |
| "valid_targets_min": 5485 | |
| }, | |
| { | |
| "epoch": 0.09264620729588882, | |
| "grad_norm": 0.533139623615456, | |
| "learning_rate": 5.223140495867769e-06, | |
| "loss": 0.3666, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16596432030200958, | |
| "step": 80, | |
| "valid_targets_mean": 7016.6, | |
| "valid_targets_min": 4960 | |
| }, | |
| { | |
| "epoch": 0.09843659525188188, | |
| "grad_norm": 0.5366119526491792, | |
| "learning_rate": 5.553719008264463e-06, | |
| "loss": 0.3524, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16567102074623108, | |
| "step": 85, | |
| "valid_targets_mean": 6527.1, | |
| "valid_targets_min": 5399 | |
| }, | |
| { | |
| "epoch": 0.10422698320787492, | |
| "grad_norm": 0.6389437092982323, | |
| "learning_rate": 5.8842975206611575e-06, | |
| "loss": 0.3516, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1750902235507965, | |
| "step": 90, | |
| "valid_targets_mean": 4995.0, | |
| "valid_targets_min": 618 | |
| }, | |
| { | |
| "epoch": 0.11001737116386798, | |
| "grad_norm": 0.5018074528544499, | |
| "learning_rate": 6.214876033057852e-06, | |
| "loss": 0.3609, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18015700578689575, | |
| "step": 95, | |
| "valid_targets_mean": 7219.4, | |
| "valid_targets_min": 5764 | |
| }, | |
| { | |
| "epoch": 0.11580775911986103, | |
| "grad_norm": 0.5081270494729966, | |
| "learning_rate": 6.545454545454546e-06, | |
| "loss": 0.3409, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17924398183822632, | |
| "step": 100, | |
| "valid_targets_mean": 7998.5, | |
| "valid_targets_min": 5902 | |
| }, | |
| { | |
| "epoch": 0.12159814707585408, | |
| "grad_norm": 0.45522046003860905, | |
| "learning_rate": 6.87603305785124e-06, | |
| "loss": 0.312, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13894960284233093, | |
| "step": 105, | |
| "valid_targets_mean": 7377.4, | |
| "valid_targets_min": 5015 | |
| }, | |
| { | |
| "epoch": 0.12738853503184713, | |
| "grad_norm": 0.4623081465603883, | |
| "learning_rate": 7.206611570247934e-06, | |
| "loss": 0.315, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1380847692489624, | |
| "step": 110, | |
| "valid_targets_mean": 6794.0, | |
| "valid_targets_min": 5283 | |
| }, | |
| { | |
| "epoch": 0.13317892298784018, | |
| "grad_norm": 0.5569607497270949, | |
| "learning_rate": 7.537190082644628e-06, | |
| "loss": 0.3279, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1848008781671524, | |
| "step": 115, | |
| "valid_targets_mean": 6656.4, | |
| "valid_targets_min": 4364 | |
| }, | |
| { | |
| "epoch": 0.13896931094383325, | |
| "grad_norm": 0.4781151946290797, | |
| "learning_rate": 7.867768595041323e-06, | |
| "loss": 0.3158, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1498945653438568, | |
| "step": 120, | |
| "valid_targets_mean": 7178.4, | |
| "valid_targets_min": 4448 | |
| }, | |
| { | |
| "epoch": 0.1447596988998263, | |
| "grad_norm": 0.48824166066226243, | |
| "learning_rate": 8.198347107438017e-06, | |
| "loss": 0.3109, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.176904559135437, | |
| "step": 125, | |
| "valid_targets_mean": 7619.8, | |
| "valid_targets_min": 5737 | |
| }, | |
| { | |
| "epoch": 0.15055008685581933, | |
| "grad_norm": 0.4260819057733041, | |
| "learning_rate": 8.528925619834712e-06, | |
| "loss": 0.2745, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1333511471748352, | |
| "step": 130, | |
| "valid_targets_mean": 8754.4, | |
| "valid_targets_min": 5752 | |
| }, | |
| { | |
| "epoch": 0.1563404748118124, | |
| "grad_norm": 0.46330713437709214, | |
| "learning_rate": 8.859504132231406e-06, | |
| "loss": 0.2729, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14825156331062317, | |
| "step": 135, | |
| "valid_targets_mean": 6341.0, | |
| "valid_targets_min": 4543 | |
| }, | |
| { | |
| "epoch": 0.16213086276780544, | |
| "grad_norm": 0.5057694929436107, | |
| "learning_rate": 9.1900826446281e-06, | |
| "loss": 0.2623, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1494828760623932, | |
| "step": 140, | |
| "valid_targets_mean": 7181.9, | |
| "valid_targets_min": 4563 | |
| }, | |
| { | |
| "epoch": 0.1679212507237985, | |
| "grad_norm": 1.0547789082478412, | |
| "learning_rate": 9.520661157024794e-06, | |
| "loss": 0.3213, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11880066245794296, | |
| "step": 145, | |
| "valid_targets_mean": 1104.4, | |
| "valid_targets_min": 161 | |
| }, | |
| { | |
| "epoch": 0.17371163867979156, | |
| "grad_norm": 0.41334982805296444, | |
| "learning_rate": 9.851239669421488e-06, | |
| "loss": 0.3041, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13268011808395386, | |
| "step": 150, | |
| "valid_targets_mean": 7018.2, | |
| "valid_targets_min": 5424 | |
| }, | |
| { | |
| "epoch": 0.1795020266357846, | |
| "grad_norm": 0.49029004832293016, | |
| "learning_rate": 1.0181818181818182e-05, | |
| "loss": 0.306, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1567586362361908, | |
| "step": 155, | |
| "valid_targets_mean": 7738.4, | |
| "valid_targets_min": 5329 | |
| }, | |
| { | |
| "epoch": 0.18529241459177764, | |
| "grad_norm": 0.47121226347035616, | |
| "learning_rate": 1.0512396694214877e-05, | |
| "loss": 0.3125, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15164563059806824, | |
| "step": 160, | |
| "valid_targets_mean": 6426.6, | |
| "valid_targets_min": 5560 | |
| }, | |
| { | |
| "epoch": 0.1910828025477707, | |
| "grad_norm": 0.5141163046607614, | |
| "learning_rate": 1.084297520661157e-05, | |
| "loss": 0.3033, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15433678030967712, | |
| "step": 165, | |
| "valid_targets_mean": 6169.9, | |
| "valid_targets_min": 3446 | |
| }, | |
| { | |
| "epoch": 0.19687319050376376, | |
| "grad_norm": 0.48615843552385, | |
| "learning_rate": 1.1173553719008265e-05, | |
| "loss": 0.2892, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1489882469177246, | |
| "step": 170, | |
| "valid_targets_mean": 7204.6, | |
| "valid_targets_min": 5381 | |
| }, | |
| { | |
| "epoch": 0.2026635784597568, | |
| "grad_norm": 0.4424369145671002, | |
| "learning_rate": 1.1504132231404959e-05, | |
| "loss": 0.2939, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15045370161533356, | |
| "step": 175, | |
| "valid_targets_mean": 7600.4, | |
| "valid_targets_min": 5241 | |
| }, | |
| { | |
| "epoch": 0.20845396641574984, | |
| "grad_norm": 0.5025657591779371, | |
| "learning_rate": 1.1834710743801653e-05, | |
| "loss": 0.285, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1559857428073883, | |
| "step": 180, | |
| "valid_targets_mean": 7098.5, | |
| "valid_targets_min": 5607 | |
| }, | |
| { | |
| "epoch": 0.2142443543717429, | |
| "grad_norm": 0.5168918509248777, | |
| "learning_rate": 1.2165289256198347e-05, | |
| "loss": 0.2902, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15138767659664154, | |
| "step": 185, | |
| "valid_targets_mean": 6955.1, | |
| "valid_targets_min": 5530 | |
| }, | |
| { | |
| "epoch": 0.22003474232773595, | |
| "grad_norm": 0.4847154459970703, | |
| "learning_rate": 1.2495867768595043e-05, | |
| "loss": 0.2847, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14534181356430054, | |
| "step": 190, | |
| "valid_targets_mean": 6748.1, | |
| "valid_targets_min": 5364 | |
| }, | |
| { | |
| "epoch": 0.225825130283729, | |
| "grad_norm": 0.5142233994213916, | |
| "learning_rate": 1.2826446280991736e-05, | |
| "loss": 0.2883, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14052218198776245, | |
| "step": 195, | |
| "valid_targets_mean": 6273.9, | |
| "valid_targets_min": 5438 | |
| }, | |
| { | |
| "epoch": 0.23161551823972207, | |
| "grad_norm": 0.46607148208472965, | |
| "learning_rate": 1.3157024793388432e-05, | |
| "loss": 0.2884, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1403881311416626, | |
| "step": 200, | |
| "valid_targets_mean": 6419.5, | |
| "valid_targets_min": 4879 | |
| }, | |
| { | |
| "epoch": 0.2374059061957151, | |
| "grad_norm": 0.46894354953159606, | |
| "learning_rate": 1.3487603305785124e-05, | |
| "loss": 0.2841, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13953427970409393, | |
| "step": 205, | |
| "valid_targets_mean": 6483.4, | |
| "valid_targets_min": 4309 | |
| }, | |
| { | |
| "epoch": 0.24319629415170815, | |
| "grad_norm": 0.5298027497127495, | |
| "learning_rate": 1.381818181818182e-05, | |
| "loss": 0.2807, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15868917107582092, | |
| "step": 210, | |
| "valid_targets_mean": 6825.0, | |
| "valid_targets_min": 5205 | |
| }, | |
| { | |
| "epoch": 0.24898668210770122, | |
| "grad_norm": 0.48339105464033155, | |
| "learning_rate": 1.4148760330578512e-05, | |
| "loss": 0.2769, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12778641283512115, | |
| "step": 215, | |
| "valid_targets_mean": 6729.4, | |
| "valid_targets_min": 5200 | |
| }, | |
| { | |
| "epoch": 0.25477707006369427, | |
| "grad_norm": 0.5273655047609594, | |
| "learning_rate": 1.4479338842975208e-05, | |
| "loss": 0.2741, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13452228903770447, | |
| "step": 220, | |
| "valid_targets_mean": 6094.0, | |
| "valid_targets_min": 5302 | |
| }, | |
| { | |
| "epoch": 0.26056745801968734, | |
| "grad_norm": 0.5545406735891124, | |
| "learning_rate": 1.48099173553719e-05, | |
| "loss": 0.2778, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1445760726928711, | |
| "step": 225, | |
| "valid_targets_mean": 7140.0, | |
| "valid_targets_min": 5270 | |
| }, | |
| { | |
| "epoch": 0.26635784597568035, | |
| "grad_norm": 0.5053786373843324, | |
| "learning_rate": 1.5140495867768596e-05, | |
| "loss": 0.2678, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12794765830039978, | |
| "step": 230, | |
| "valid_targets_mean": 7425.0, | |
| "valid_targets_min": 5281 | |
| }, | |
| { | |
| "epoch": 0.2721482339316734, | |
| "grad_norm": 0.5117754491840001, | |
| "learning_rate": 1.547107438016529e-05, | |
| "loss": 0.2748, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1676265448331833, | |
| "step": 235, | |
| "valid_targets_mean": 7063.4, | |
| "valid_targets_min": 5386 | |
| }, | |
| { | |
| "epoch": 0.2779386218876665, | |
| "grad_norm": 0.4967372367737128, | |
| "learning_rate": 1.5801652892561985e-05, | |
| "loss": 0.2744, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.144246906042099, | |
| "step": 240, | |
| "valid_targets_mean": 6827.0, | |
| "valid_targets_min": 4547 | |
| }, | |
| { | |
| "epoch": 0.2837290098436595, | |
| "grad_norm": 0.7973056364596196, | |
| "learning_rate": 1.613223140495868e-05, | |
| "loss": 0.2538, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08753351122140884, | |
| "step": 245, | |
| "valid_targets_mean": 2047.0, | |
| "valid_targets_min": 165 | |
| }, | |
| { | |
| "epoch": 0.2895193977996526, | |
| "grad_norm": 0.4943520816825678, | |
| "learning_rate": 1.6462809917355373e-05, | |
| "loss": 0.2642, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13315731287002563, | |
| "step": 250, | |
| "valid_targets_mean": 6754.1, | |
| "valid_targets_min": 5196 | |
| }, | |
| { | |
| "epoch": 0.29530978575564565, | |
| "grad_norm": 0.5498952539845848, | |
| "learning_rate": 1.6793388429752067e-05, | |
| "loss": 0.2649, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13404254615306854, | |
| "step": 255, | |
| "valid_targets_mean": 5985.6, | |
| "valid_targets_min": 5028 | |
| }, | |
| { | |
| "epoch": 0.30110017371163866, | |
| "grad_norm": 0.5479083359283229, | |
| "learning_rate": 1.712396694214876e-05, | |
| "loss": 0.2699, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1222841888666153, | |
| "step": 260, | |
| "valid_targets_mean": 5963.9, | |
| "valid_targets_min": 5237 | |
| }, | |
| { | |
| "epoch": 0.30689056166763173, | |
| "grad_norm": 0.5438335340635813, | |
| "learning_rate": 1.7454545454545456e-05, | |
| "loss": 0.2644, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13820022344589233, | |
| "step": 265, | |
| "valid_targets_mean": 6319.0, | |
| "valid_targets_min": 5412 | |
| }, | |
| { | |
| "epoch": 0.3126809496236248, | |
| "grad_norm": 0.4652892213452802, | |
| "learning_rate": 1.778512396694215e-05, | |
| "loss": 0.2506, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1078779399394989, | |
| "step": 270, | |
| "valid_targets_mean": 7125.8, | |
| "valid_targets_min": 5080 | |
| }, | |
| { | |
| "epoch": 0.3184713375796178, | |
| "grad_norm": 0.5103127506468167, | |
| "learning_rate": 1.8115702479338844e-05, | |
| "loss": 0.2648, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1312594711780548, | |
| "step": 275, | |
| "valid_targets_mean": 6123.5, | |
| "valid_targets_min": 4521 | |
| }, | |
| { | |
| "epoch": 0.3242617255356109, | |
| "grad_norm": 1.2150630310249255, | |
| "learning_rate": 1.8446280991735538e-05, | |
| "loss": 0.3456, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22883066534996033, | |
| "step": 280, | |
| "valid_targets_mean": 6020.6, | |
| "valid_targets_min": 2493 | |
| }, | |
| { | |
| "epoch": 0.33005211349160396, | |
| "grad_norm": 0.7876446225017467, | |
| "learning_rate": 1.8776859504132232e-05, | |
| "loss": 0.426, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.171472430229187, | |
| "step": 285, | |
| "valid_targets_mean": 3953.5, | |
| "valid_targets_min": 1177 | |
| }, | |
| { | |
| "epoch": 0.335842501447597, | |
| "grad_norm": 0.735005861492704, | |
| "learning_rate": 1.9107438016528926e-05, | |
| "loss": 0.4201, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24447613954544067, | |
| "step": 290, | |
| "valid_targets_mean": 4950.8, | |
| "valid_targets_min": 2799 | |
| }, | |
| { | |
| "epoch": 0.34163288940359005, | |
| "grad_norm": 0.7105988732526015, | |
| "learning_rate": 1.943801652892562e-05, | |
| "loss": 0.3965, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2215518057346344, | |
| "step": 295, | |
| "valid_targets_mean": 5312.0, | |
| "valid_targets_min": 2973 | |
| }, | |
| { | |
| "epoch": 0.3474232773595831, | |
| "grad_norm": 0.6727827921481346, | |
| "learning_rate": 1.9768595041322315e-05, | |
| "loss": 0.3889, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18209363520145416, | |
| "step": 300, | |
| "valid_targets_mean": 3775.5, | |
| "valid_targets_min": 980 | |
| }, | |
| { | |
| "epoch": 0.35321366531557613, | |
| "grad_norm": 0.6668317415236219, | |
| "learning_rate": 2.0099173553719012e-05, | |
| "loss": 0.4069, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20714129507541656, | |
| "step": 305, | |
| "valid_targets_mean": 5055.4, | |
| "valid_targets_min": 1512 | |
| }, | |
| { | |
| "epoch": 0.3590040532715692, | |
| "grad_norm": 0.6107245060489721, | |
| "learning_rate": 2.0429752066115703e-05, | |
| "loss": 0.374, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14312531054019928, | |
| "step": 310, | |
| "valid_targets_mean": 3600.8, | |
| "valid_targets_min": 1848 | |
| }, | |
| { | |
| "epoch": 0.36479444122756227, | |
| "grad_norm": 0.5509653204998359, | |
| "learning_rate": 2.0760330578512397e-05, | |
| "loss": 0.3772, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13793471455574036, | |
| "step": 315, | |
| "valid_targets_mean": 3369.8, | |
| "valid_targets_min": 1381 | |
| }, | |
| { | |
| "epoch": 0.3705848291835553, | |
| "grad_norm": 0.5293595320233865, | |
| "learning_rate": 2.109090909090909e-05, | |
| "loss": 0.3804, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19333019852638245, | |
| "step": 320, | |
| "valid_targets_mean": 6283.2, | |
| "valid_targets_min": 3600 | |
| }, | |
| { | |
| "epoch": 0.37637521713954836, | |
| "grad_norm": 0.5779763292618605, | |
| "learning_rate": 2.142148760330579e-05, | |
| "loss": 0.3946, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19307959079742432, | |
| "step": 325, | |
| "valid_targets_mean": 5242.5, | |
| "valid_targets_min": 2540 | |
| }, | |
| { | |
| "epoch": 0.3821656050955414, | |
| "grad_norm": 0.5507055471870357, | |
| "learning_rate": 2.1752066115702483e-05, | |
| "loss": 0.376, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17400136590003967, | |
| "step": 330, | |
| "valid_targets_mean": 6246.5, | |
| "valid_targets_min": 2157 | |
| }, | |
| { | |
| "epoch": 0.38795599305153444, | |
| "grad_norm": 0.6129232711123003, | |
| "learning_rate": 2.2082644628099174e-05, | |
| "loss": 0.376, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19069461524486542, | |
| "step": 335, | |
| "valid_targets_mean": 4556.5, | |
| "valid_targets_min": 1720 | |
| }, | |
| { | |
| "epoch": 0.3937463810075275, | |
| "grad_norm": 0.6201979514739216, | |
| "learning_rate": 2.2413223140495868e-05, | |
| "loss": 0.3798, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21241924166679382, | |
| "step": 340, | |
| "valid_targets_mean": 5116.5, | |
| "valid_targets_min": 1982 | |
| }, | |
| { | |
| "epoch": 0.3995367689635206, | |
| "grad_norm": 0.6030358543858219, | |
| "learning_rate": 2.2743801652892566e-05, | |
| "loss": 0.3714, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.188879132270813, | |
| "step": 345, | |
| "valid_targets_mean": 4464.5, | |
| "valid_targets_min": 2718 | |
| }, | |
| { | |
| "epoch": 0.4053271569195136, | |
| "grad_norm": 0.6841676419341239, | |
| "learning_rate": 2.307438016528926e-05, | |
| "loss": 0.3853, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1824164092540741, | |
| "step": 350, | |
| "valid_targets_mean": 3418.5, | |
| "valid_targets_min": 1451 | |
| }, | |
| { | |
| "epoch": 0.41111754487550667, | |
| "grad_norm": 0.6305854851252951, | |
| "learning_rate": 2.340495867768595e-05, | |
| "loss": 0.3829, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1953430473804474, | |
| "step": 355, | |
| "valid_targets_mean": 3990.4, | |
| "valid_targets_min": 1861 | |
| }, | |
| { | |
| "epoch": 0.4169079328314997, | |
| "grad_norm": 0.6879145519243186, | |
| "learning_rate": 2.3735537190082645e-05, | |
| "loss": 0.3742, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2048986852169037, | |
| "step": 360, | |
| "valid_targets_mean": 4065.4, | |
| "valid_targets_min": 2100 | |
| }, | |
| { | |
| "epoch": 0.42269832078749275, | |
| "grad_norm": 0.670617667592637, | |
| "learning_rate": 2.4066115702479342e-05, | |
| "loss": 0.3864, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21338799595832825, | |
| "step": 365, | |
| "valid_targets_mean": 4132.4, | |
| "valid_targets_min": 3234 | |
| }, | |
| { | |
| "epoch": 0.4284887087434858, | |
| "grad_norm": 0.7359672228538012, | |
| "learning_rate": 2.4396694214876036e-05, | |
| "loss": 0.3932, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23431096971035004, | |
| "step": 370, | |
| "valid_targets_mean": 4246.2, | |
| "valid_targets_min": 2650 | |
| }, | |
| { | |
| "epoch": 0.43427909669947884, | |
| "grad_norm": 0.7503953347940379, | |
| "learning_rate": 2.4727272727272727e-05, | |
| "loss": 0.3808, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1593906581401825, | |
| "step": 375, | |
| "valid_targets_mean": 4096.5, | |
| "valid_targets_min": 1719 | |
| }, | |
| { | |
| "epoch": 0.4400694846554719, | |
| "grad_norm": 0.696019831669972, | |
| "learning_rate": 2.505785123966942e-05, | |
| "loss": 0.3851, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18997083604335785, | |
| "step": 380, | |
| "valid_targets_mean": 3933.1, | |
| "valid_targets_min": 1020 | |
| }, | |
| { | |
| "epoch": 0.445859872611465, | |
| "grad_norm": 1.0940174628721868, | |
| "learning_rate": 2.538842975206612e-05, | |
| "loss": 0.3627, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18985775113105774, | |
| "step": 385, | |
| "valid_targets_mean": 3551.1, | |
| "valid_targets_min": 2130 | |
| }, | |
| { | |
| "epoch": 0.451650260567458, | |
| "grad_norm": 0.6578172297344157, | |
| "learning_rate": 2.5719008264462813e-05, | |
| "loss": 0.3556, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15262863039970398, | |
| "step": 390, | |
| "valid_targets_mean": 3535.0, | |
| "valid_targets_min": 865 | |
| }, | |
| { | |
| "epoch": 0.45744064852345107, | |
| "grad_norm": 0.9405521225813999, | |
| "learning_rate": 2.6049586776859507e-05, | |
| "loss": 0.3656, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.248321533203125, | |
| "step": 395, | |
| "valid_targets_mean": 3716.5, | |
| "valid_targets_min": 3144 | |
| }, | |
| { | |
| "epoch": 0.46323103647944414, | |
| "grad_norm": 1.0432523503468756, | |
| "learning_rate": 2.6380165289256198e-05, | |
| "loss": 0.3739, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1557130515575409, | |
| "step": 400, | |
| "valid_targets_mean": 3262.2, | |
| "valid_targets_min": 2050 | |
| }, | |
| { | |
| "epoch": 0.46902142443543715, | |
| "grad_norm": 0.6632192119125181, | |
| "learning_rate": 2.6710743801652895e-05, | |
| "loss": 0.372, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18791770935058594, | |
| "step": 405, | |
| "valid_targets_mean": 3296.5, | |
| "valid_targets_min": 1570 | |
| }, | |
| { | |
| "epoch": 0.4748118123914302, | |
| "grad_norm": 0.6008101768878966, | |
| "learning_rate": 2.704132231404959e-05, | |
| "loss": 0.353, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17852509021759033, | |
| "step": 410, | |
| "valid_targets_mean": 4806.4, | |
| "valid_targets_min": 1795 | |
| }, | |
| { | |
| "epoch": 0.4806022003474233, | |
| "grad_norm": 0.6184314563964771, | |
| "learning_rate": 2.7371900826446284e-05, | |
| "loss": 0.3566, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16312438249588013, | |
| "step": 415, | |
| "valid_targets_mean": 4342.4, | |
| "valid_targets_min": 1934 | |
| }, | |
| { | |
| "epoch": 0.4863925883034163, | |
| "grad_norm": 0.6917524383618637, | |
| "learning_rate": 2.7702479338842974e-05, | |
| "loss": 0.3666, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18566149473190308, | |
| "step": 420, | |
| "valid_targets_mean": 3675.4, | |
| "valid_targets_min": 2557 | |
| }, | |
| { | |
| "epoch": 0.4921829762594094, | |
| "grad_norm": 0.6379702051172501, | |
| "learning_rate": 2.8033057851239672e-05, | |
| "loss": 0.3625, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20107987523078918, | |
| "step": 425, | |
| "valid_targets_mean": 4550.4, | |
| "valid_targets_min": 2663 | |
| }, | |
| { | |
| "epoch": 0.49797336421540245, | |
| "grad_norm": 0.6626061187107526, | |
| "learning_rate": 2.8363636363636366e-05, | |
| "loss": 0.3522, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1774175465106964, | |
| "step": 430, | |
| "valid_targets_mean": 4604.0, | |
| "valid_targets_min": 2016 | |
| }, | |
| { | |
| "epoch": 0.5037637521713955, | |
| "grad_norm": 0.703691101341157, | |
| "learning_rate": 2.869421487603306e-05, | |
| "loss": 0.3598, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15914887189865112, | |
| "step": 435, | |
| "valid_targets_mean": 2976.8, | |
| "valid_targets_min": 1367 | |
| }, | |
| { | |
| "epoch": 0.5095541401273885, | |
| "grad_norm": 0.7067311357342676, | |
| "learning_rate": 2.902479338842975e-05, | |
| "loss": 0.3676, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17488998174667358, | |
| "step": 440, | |
| "valid_targets_mean": 3542.5, | |
| "valid_targets_min": 2026 | |
| }, | |
| { | |
| "epoch": 0.5153445280833816, | |
| "grad_norm": 0.6737400104351488, | |
| "learning_rate": 2.9355371900826452e-05, | |
| "loss": 0.3475, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16618910431861877, | |
| "step": 445, | |
| "valid_targets_mean": 3247.2, | |
| "valid_targets_min": 2033 | |
| }, | |
| { | |
| "epoch": 0.5211349160393747, | |
| "grad_norm": 0.8326243547681798, | |
| "learning_rate": 2.9685950413223143e-05, | |
| "loss": 0.3699, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1832060068845749, | |
| "step": 450, | |
| "valid_targets_mean": 2201.6, | |
| "valid_targets_min": 1283 | |
| }, | |
| { | |
| "epoch": 0.5269253039953677, | |
| "grad_norm": 0.8234864634523447, | |
| "learning_rate": 3.0016528925619837e-05, | |
| "loss": 0.3616, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18912853300571442, | |
| "step": 455, | |
| "valid_targets_mean": 2824.8, | |
| "valid_targets_min": 1636 | |
| }, | |
| { | |
| "epoch": 0.5327156919513607, | |
| "grad_norm": 0.7135386777132873, | |
| "learning_rate": 3.034710743801653e-05, | |
| "loss": 0.3596, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1965169608592987, | |
| "step": 460, | |
| "valid_targets_mean": 3844.4, | |
| "valid_targets_min": 1629 | |
| }, | |
| { | |
| "epoch": 0.5385060799073538, | |
| "grad_norm": 0.7315405541721403, | |
| "learning_rate": 3.067768595041323e-05, | |
| "loss": 0.3494, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16616566479206085, | |
| "step": 465, | |
| "valid_targets_mean": 3048.6, | |
| "valid_targets_min": 2080 | |
| }, | |
| { | |
| "epoch": 0.5442964678633468, | |
| "grad_norm": 0.6925498446822687, | |
| "learning_rate": 3.100826446280992e-05, | |
| "loss": 0.3542, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16240294277668, | |
| "step": 470, | |
| "valid_targets_mean": 3111.8, | |
| "valid_targets_min": 1153 | |
| }, | |
| { | |
| "epoch": 0.5500868558193399, | |
| "grad_norm": 0.6353538526882198, | |
| "learning_rate": 3.133884297520662e-05, | |
| "loss": 0.3476, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19012531638145447, | |
| "step": 475, | |
| "valid_targets_mean": 4069.0, | |
| "valid_targets_min": 1858 | |
| }, | |
| { | |
| "epoch": 0.555877243775333, | |
| "grad_norm": 0.7238753552976274, | |
| "learning_rate": 3.166942148760331e-05, | |
| "loss": 0.3639, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15231537818908691, | |
| "step": 480, | |
| "valid_targets_mean": 3415.2, | |
| "valid_targets_min": 1639 | |
| }, | |
| { | |
| "epoch": 0.561667631731326, | |
| "grad_norm": 0.6665274482956017, | |
| "learning_rate": 3.2000000000000005e-05, | |
| "loss": 0.3569, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.145134836435318, | |
| "step": 485, | |
| "valid_targets_mean": 3267.6, | |
| "valid_targets_min": 1394 | |
| }, | |
| { | |
| "epoch": 0.567458019687319, | |
| "grad_norm": 0.6510657131966783, | |
| "learning_rate": 3.2330578512396696e-05, | |
| "loss": 0.3484, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18342448770999908, | |
| "step": 490, | |
| "valid_targets_mean": 4775.9, | |
| "valid_targets_min": 1844 | |
| }, | |
| { | |
| "epoch": 0.5732484076433121, | |
| "grad_norm": 1.018435315706297, | |
| "learning_rate": 3.2661157024793394e-05, | |
| "loss": 0.3508, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15765246748924255, | |
| "step": 495, | |
| "valid_targets_mean": 3317.1, | |
| "valid_targets_min": 1305 | |
| }, | |
| { | |
| "epoch": 0.5790387955993052, | |
| "grad_norm": 0.6883157544394269, | |
| "learning_rate": 3.2991735537190084e-05, | |
| "loss": 0.3468, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1881387084722519, | |
| "step": 500, | |
| "valid_targets_mean": 4290.8, | |
| "valid_targets_min": 2185 | |
| }, | |
| { | |
| "epoch": 0.5848291835552982, | |
| "grad_norm": 0.6721803569935538, | |
| "learning_rate": 3.332231404958678e-05, | |
| "loss": 0.3537, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20651760697364807, | |
| "step": 505, | |
| "valid_targets_mean": 4024.9, | |
| "valid_targets_min": 2349 | |
| }, | |
| { | |
| "epoch": 0.5906195715112913, | |
| "grad_norm": 0.6291206567565695, | |
| "learning_rate": 3.365289256198347e-05, | |
| "loss": 0.3359, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15432091057300568, | |
| "step": 510, | |
| "valid_targets_mean": 3453.0, | |
| "valid_targets_min": 1607 | |
| }, | |
| { | |
| "epoch": 0.5964099594672843, | |
| "grad_norm": 0.6904459046779862, | |
| "learning_rate": 3.398347107438017e-05, | |
| "loss": 0.362, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1796693503856659, | |
| "step": 515, | |
| "valid_targets_mean": 3389.8, | |
| "valid_targets_min": 1507 | |
| }, | |
| { | |
| "epoch": 0.6022003474232773, | |
| "grad_norm": 0.640463878413541, | |
| "learning_rate": 3.431404958677686e-05, | |
| "loss": 0.357, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16405197978019714, | |
| "step": 520, | |
| "valid_targets_mean": 3773.9, | |
| "valid_targets_min": 1317 | |
| }, | |
| { | |
| "epoch": 0.6079907353792704, | |
| "grad_norm": 0.6511127283978864, | |
| "learning_rate": 3.464462809917356e-05, | |
| "loss": 0.3493, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15977990627288818, | |
| "step": 525, | |
| "valid_targets_mean": 3881.0, | |
| "valid_targets_min": 3355 | |
| }, | |
| { | |
| "epoch": 0.6137811233352635, | |
| "grad_norm": 0.6341328479062885, | |
| "learning_rate": 3.497520661157025e-05, | |
| "loss": 0.3506, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16847799718379974, | |
| "step": 530, | |
| "valid_targets_mean": 3951.9, | |
| "valid_targets_min": 2999 | |
| }, | |
| { | |
| "epoch": 0.6195715112912565, | |
| "grad_norm": 0.6419289214078541, | |
| "learning_rate": 3.530578512396695e-05, | |
| "loss": 0.35, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15617629885673523, | |
| "step": 535, | |
| "valid_targets_mean": 4086.5, | |
| "valid_targets_min": 1739 | |
| }, | |
| { | |
| "epoch": 0.6253618992472496, | |
| "grad_norm": 0.6420357082141755, | |
| "learning_rate": 3.563636363636364e-05, | |
| "loss": 0.3502, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15255317091941833, | |
| "step": 540, | |
| "valid_targets_mean": 3298.0, | |
| "valid_targets_min": 1140 | |
| }, | |
| { | |
| "epoch": 0.6311522872032426, | |
| "grad_norm": 0.6147364203602501, | |
| "learning_rate": 3.5966942148760335e-05, | |
| "loss": 0.3288, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18115392327308655, | |
| "step": 545, | |
| "valid_targets_mean": 4012.1, | |
| "valid_targets_min": 2779 | |
| }, | |
| { | |
| "epoch": 0.6369426751592356, | |
| "grad_norm": 0.6415078707723959, | |
| "learning_rate": 3.6297520661157026e-05, | |
| "loss": 0.3405, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16753660142421722, | |
| "step": 550, | |
| "valid_targets_mean": 3636.9, | |
| "valid_targets_min": 1183 | |
| }, | |
| { | |
| "epoch": 0.6427330631152287, | |
| "grad_norm": 0.6598650414716416, | |
| "learning_rate": 3.6628099173553724e-05, | |
| "loss": 0.3262, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19901029765605927, | |
| "step": 555, | |
| "valid_targets_mean": 4269.0, | |
| "valid_targets_min": 3041 | |
| }, | |
| { | |
| "epoch": 0.6485234510712218, | |
| "grad_norm": 0.6794989569457508, | |
| "learning_rate": 3.6958677685950414e-05, | |
| "loss": 0.3568, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17811614274978638, | |
| "step": 560, | |
| "valid_targets_mean": 3350.1, | |
| "valid_targets_min": 1726 | |
| }, | |
| { | |
| "epoch": 0.6543138390272148, | |
| "grad_norm": 0.5301401186579153, | |
| "learning_rate": 3.728925619834711e-05, | |
| "loss": 0.3349, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15499885380268097, | |
| "step": 565, | |
| "valid_targets_mean": 4747.4, | |
| "valid_targets_min": 1743 | |
| }, | |
| { | |
| "epoch": 0.6601042269832079, | |
| "grad_norm": 0.6146825808496802, | |
| "learning_rate": 3.76198347107438e-05, | |
| "loss": 0.347, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16681280732154846, | |
| "step": 570, | |
| "valid_targets_mean": 3702.8, | |
| "valid_targets_min": 1234 | |
| }, | |
| { | |
| "epoch": 0.6658946149392009, | |
| "grad_norm": 0.6199338035415232, | |
| "learning_rate": 3.79504132231405e-05, | |
| "loss": 0.3533, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17797984182834625, | |
| "step": 575, | |
| "valid_targets_mean": 4079.0, | |
| "valid_targets_min": 2303 | |
| }, | |
| { | |
| "epoch": 0.671685002895194, | |
| "grad_norm": 0.6414527008678907, | |
| "learning_rate": 3.828099173553719e-05, | |
| "loss": 0.353, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17539098858833313, | |
| "step": 580, | |
| "valid_targets_mean": 3830.8, | |
| "valid_targets_min": 1651 | |
| }, | |
| { | |
| "epoch": 0.677475390851187, | |
| "grad_norm": 0.7158891749544349, | |
| "learning_rate": 3.861157024793389e-05, | |
| "loss": 0.3437, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.184058278799057, | |
| "step": 585, | |
| "valid_targets_mean": 2876.2, | |
| "valid_targets_min": 1374 | |
| }, | |
| { | |
| "epoch": 0.6832657788071801, | |
| "grad_norm": 0.725484792932723, | |
| "learning_rate": 3.894214876033058e-05, | |
| "loss": 0.3347, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16789329051971436, | |
| "step": 590, | |
| "valid_targets_mean": 3467.5, | |
| "valid_targets_min": 1625 | |
| }, | |
| { | |
| "epoch": 0.6890561667631732, | |
| "grad_norm": 0.6355075281081544, | |
| "learning_rate": 3.927272727272728e-05, | |
| "loss": 0.3299, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1683839112520218, | |
| "step": 595, | |
| "valid_targets_mean": 4089.5, | |
| "valid_targets_min": 997 | |
| }, | |
| { | |
| "epoch": 0.6948465547191662, | |
| "grad_norm": 0.5717834677186259, | |
| "learning_rate": 3.960330578512397e-05, | |
| "loss": 0.3363, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14123469591140747, | |
| "step": 600, | |
| "valid_targets_mean": 3618.6, | |
| "valid_targets_min": 1179 | |
| }, | |
| { | |
| "epoch": 0.7006369426751592, | |
| "grad_norm": 0.7698731162915861, | |
| "learning_rate": 3.9933884297520665e-05, | |
| "loss": 0.3348, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16036216914653778, | |
| "step": 605, | |
| "valid_targets_mean": 3079.1, | |
| "valid_targets_min": 1444 | |
| }, | |
| { | |
| "epoch": 0.7064273306311523, | |
| "grad_norm": 0.6676617529688202, | |
| "learning_rate": 3.9999946698078364e-05, | |
| "loss": 0.317, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1451885998249054, | |
| "step": 610, | |
| "valid_targets_mean": 3486.8, | |
| "valid_targets_min": 1504 | |
| }, | |
| { | |
| "epoch": 0.7122177185871453, | |
| "grad_norm": 0.603977760826595, | |
| "learning_rate": 3.9999730159508614e-05, | |
| "loss": 0.3293, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16658622026443481, | |
| "step": 615, | |
| "valid_targets_mean": 3998.4, | |
| "valid_targets_min": 2302 | |
| }, | |
| { | |
| "epoch": 0.7180081065431384, | |
| "grad_norm": 0.8047137177662046, | |
| "learning_rate": 3.9999347054722696e-05, | |
| "loss": 0.3392, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21027225255966187, | |
| "step": 620, | |
| "valid_targets_mean": 4282.5, | |
| "valid_targets_min": 1818 | |
| }, | |
| { | |
| "epoch": 0.7237984944991315, | |
| "grad_norm": 0.605583750962231, | |
| "learning_rate": 3.999879738691128e-05, | |
| "loss": 0.3611, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17750391364097595, | |
| "step": 625, | |
| "valid_targets_mean": 4221.9, | |
| "valid_targets_min": 2633 | |
| }, | |
| { | |
| "epoch": 0.7295888824551245, | |
| "grad_norm": 0.6129538675494742, | |
| "learning_rate": 3.999808116065221e-05, | |
| "loss": 0.3326, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17118681967258453, | |
| "step": 630, | |
| "valid_targets_mean": 3522.4, | |
| "valid_targets_min": 2481 | |
| }, | |
| { | |
| "epoch": 0.7353792704111175, | |
| "grad_norm": 0.6501310721805252, | |
| "learning_rate": 3.999719838191054e-05, | |
| "loss": 0.3458, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18486127257347107, | |
| "step": 635, | |
| "valid_targets_mean": 4058.8, | |
| "valid_targets_min": 2013 | |
| }, | |
| { | |
| "epoch": 0.7411696583671106, | |
| "grad_norm": 0.7104788952748324, | |
| "learning_rate": 3.999614905803841e-05, | |
| "loss": 0.3269, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17827048897743225, | |
| "step": 640, | |
| "valid_targets_mean": 3723.5, | |
| "valid_targets_min": 1744 | |
| }, | |
| { | |
| "epoch": 0.7469600463231036, | |
| "grad_norm": 0.622483649564905, | |
| "learning_rate": 3.999493319777504e-05, | |
| "loss": 0.3418, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1791311353445053, | |
| "step": 645, | |
| "valid_targets_mean": 4180.9, | |
| "valid_targets_min": 1914 | |
| }, | |
| { | |
| "epoch": 0.7527504342790967, | |
| "grad_norm": 0.6633582586829129, | |
| "learning_rate": 3.999355081124663e-05, | |
| "loss": 0.3269, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1443580985069275, | |
| "step": 650, | |
| "valid_targets_mean": 3668.6, | |
| "valid_targets_min": 3065 | |
| }, | |
| { | |
| "epoch": 0.7585408222350898, | |
| "grad_norm": 0.5827334066652042, | |
| "learning_rate": 3.9992001909966284e-05, | |
| "loss": 0.3169, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1897888481616974, | |
| "step": 655, | |
| "valid_targets_mean": 4473.2, | |
| "valid_targets_min": 2894 | |
| }, | |
| { | |
| "epoch": 0.7643312101910829, | |
| "grad_norm": 0.7444083935272425, | |
| "learning_rate": 3.99902865068339e-05, | |
| "loss": 0.3426, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17201346158981323, | |
| "step": 660, | |
| "valid_targets_mean": 3395.6, | |
| "valid_targets_min": 2359 | |
| }, | |
| { | |
| "epoch": 0.7701215981470758, | |
| "grad_norm": 0.7066890557943601, | |
| "learning_rate": 3.998840461613608e-05, | |
| "loss": 0.3302, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14887091517448425, | |
| "step": 665, | |
| "valid_targets_mean": 3114.5, | |
| "valid_targets_min": 1463 | |
| }, | |
| { | |
| "epoch": 0.7759119861030689, | |
| "grad_norm": 0.6192222424391265, | |
| "learning_rate": 3.998635625354602e-05, | |
| "loss": 0.3415, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16993939876556396, | |
| "step": 670, | |
| "valid_targets_mean": 3588.1, | |
| "valid_targets_min": 1426 | |
| }, | |
| { | |
| "epoch": 0.781702374059062, | |
| "grad_norm": 0.6542474820446244, | |
| "learning_rate": 3.998414143612333e-05, | |
| "loss": 0.3466, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16761505603790283, | |
| "step": 675, | |
| "valid_targets_mean": 3405.5, | |
| "valid_targets_min": 1661 | |
| }, | |
| { | |
| "epoch": 0.787492762015055, | |
| "grad_norm": 0.5799377222940094, | |
| "learning_rate": 3.9981760182313957e-05, | |
| "loss": 0.3365, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17293000221252441, | |
| "step": 680, | |
| "valid_targets_mean": 4457.8, | |
| "valid_targets_min": 1529 | |
| }, | |
| { | |
| "epoch": 0.7932831499710481, | |
| "grad_norm": 0.8486757316641838, | |
| "learning_rate": 3.9979212511949984e-05, | |
| "loss": 0.3693, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19929572939872742, | |
| "step": 685, | |
| "valid_targets_mean": 3272.9, | |
| "valid_targets_min": 1450 | |
| }, | |
| { | |
| "epoch": 0.7990735379270412, | |
| "grad_norm": 0.6591616098189284, | |
| "learning_rate": 3.997649844624951e-05, | |
| "loss": 0.3352, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14856567978858948, | |
| "step": 690, | |
| "valid_targets_mean": 3103.2, | |
| "valid_targets_min": 1625 | |
| }, | |
| { | |
| "epoch": 0.8048639258830341, | |
| "grad_norm": 0.5972544991840065, | |
| "learning_rate": 3.9973618007816395e-05, | |
| "loss": 0.3427, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1420861929655075, | |
| "step": 695, | |
| "valid_targets_mean": 3952.2, | |
| "valid_targets_min": 1174 | |
| }, | |
| { | |
| "epoch": 0.8106543138390272, | |
| "grad_norm": 0.5926035572764118, | |
| "learning_rate": 3.9970571220640164e-05, | |
| "loss": 0.3466, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17784333229064941, | |
| "step": 700, | |
| "valid_targets_mean": 4039.2, | |
| "valid_targets_min": 2882 | |
| }, | |
| { | |
| "epoch": 0.8164447017950203, | |
| "grad_norm": 0.7172461283418334, | |
| "learning_rate": 3.996735811009575e-05, | |
| "loss": 0.3482, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19199448823928833, | |
| "step": 705, | |
| "valid_targets_mean": 4594.0, | |
| "valid_targets_min": 1946 | |
| }, | |
| { | |
| "epoch": 0.8222350897510133, | |
| "grad_norm": 0.587560988982399, | |
| "learning_rate": 3.996397870294329e-05, | |
| "loss": 0.3385, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16461029648780823, | |
| "step": 710, | |
| "valid_targets_mean": 4082.6, | |
| "valid_targets_min": 2000 | |
| }, | |
| { | |
| "epoch": 0.8280254777070064, | |
| "grad_norm": 0.6171659688682881, | |
| "learning_rate": 3.996043302732792e-05, | |
| "loss": 0.3458, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18425926566123962, | |
| "step": 715, | |
| "valid_targets_mean": 4596.1, | |
| "valid_targets_min": 2012 | |
| }, | |
| { | |
| "epoch": 0.8338158656629994, | |
| "grad_norm": 0.6557809627187827, | |
| "learning_rate": 3.995672111277953e-05, | |
| "loss": 0.345, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17367544770240784, | |
| "step": 720, | |
| "valid_targets_mean": 3094.4, | |
| "valid_targets_min": 719 | |
| }, | |
| { | |
| "epoch": 0.8396062536189924, | |
| "grad_norm": 0.5612740562619445, | |
| "learning_rate": 3.995284299021251e-05, | |
| "loss": 0.3383, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1546105444431305, | |
| "step": 725, | |
| "valid_targets_mean": 4496.2, | |
| "valid_targets_min": 3238 | |
| }, | |
| { | |
| "epoch": 0.8453966415749855, | |
| "grad_norm": 0.6423654246977016, | |
| "learning_rate": 3.994879869192552e-05, | |
| "loss": 0.3242, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14627155661582947, | |
| "step": 730, | |
| "valid_targets_mean": 3816.8, | |
| "valid_targets_min": 1372 | |
| }, | |
| { | |
| "epoch": 0.8511870295309786, | |
| "grad_norm": 0.6898680259078053, | |
| "learning_rate": 3.994458825160117e-05, | |
| "loss": 0.3208, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1782332956790924, | |
| "step": 735, | |
| "valid_targets_mean": 3306.1, | |
| "valid_targets_min": 2370 | |
| }, | |
| { | |
| "epoch": 0.8569774174869716, | |
| "grad_norm": 0.6148055706863383, | |
| "learning_rate": 3.99402117043058e-05, | |
| "loss": 0.3326, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14432159066200256, | |
| "step": 740, | |
| "valid_targets_mean": 2958.1, | |
| "valid_targets_min": 1576 | |
| }, | |
| { | |
| "epoch": 0.8627678054429647, | |
| "grad_norm": 0.655012125079548, | |
| "learning_rate": 3.993566908648914e-05, | |
| "loss": 0.3254, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17978647351264954, | |
| "step": 745, | |
| "valid_targets_mean": 3758.5, | |
| "valid_targets_min": 1426 | |
| }, | |
| { | |
| "epoch": 0.8685581933989577, | |
| "grad_norm": 0.7067281742494269, | |
| "learning_rate": 3.993096043598403e-05, | |
| "loss": 0.3398, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14911097288131714, | |
| "step": 750, | |
| "valid_targets_mean": 3117.0, | |
| "valid_targets_min": 1728 | |
| }, | |
| { | |
| "epoch": 0.8743485813549507, | |
| "grad_norm": 0.8759168561210263, | |
| "learning_rate": 3.992608579200612e-05, | |
| "loss": 0.3339, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17803344130516052, | |
| "step": 755, | |
| "valid_targets_mean": 3264.2, | |
| "valid_targets_min": 1362 | |
| }, | |
| { | |
| "epoch": 0.8801389693109438, | |
| "grad_norm": 0.5534589430756651, | |
| "learning_rate": 3.9921045195153493e-05, | |
| "loss": 0.3313, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1561286449432373, | |
| "step": 760, | |
| "valid_targets_mean": 3836.0, | |
| "valid_targets_min": 852 | |
| }, | |
| { | |
| "epoch": 0.8859293572669369, | |
| "grad_norm": 1.3061158563859285, | |
| "learning_rate": 3.991583868740638e-05, | |
| "loss": 0.3408, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21102529764175415, | |
| "step": 765, | |
| "valid_targets_mean": 2764.1, | |
| "valid_targets_min": 1195 | |
| }, | |
| { | |
| "epoch": 0.89171974522293, | |
| "grad_norm": 0.5613147833819194, | |
| "learning_rate": 3.9910466312126786e-05, | |
| "loss": 0.3265, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17475545406341553, | |
| "step": 770, | |
| "valid_targets_mean": 4457.5, | |
| "valid_targets_min": 1564 | |
| }, | |
| { | |
| "epoch": 0.897510133178923, | |
| "grad_norm": 0.8717166999384921, | |
| "learning_rate": 3.990492811405811e-05, | |
| "loss": 0.3215, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13619567453861237, | |
| "step": 775, | |
| "valid_targets_mean": 4371.2, | |
| "valid_targets_min": 1975 | |
| }, | |
| { | |
| "epoch": 0.903300521134916, | |
| "grad_norm": 0.6096546708962742, | |
| "learning_rate": 3.989922413932482e-05, | |
| "loss": 0.3137, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16569563746452332, | |
| "step": 780, | |
| "valid_targets_mean": 3965.8, | |
| "valid_targets_min": 2426 | |
| }, | |
| { | |
| "epoch": 0.9090909090909091, | |
| "grad_norm": 0.5149683718364065, | |
| "learning_rate": 3.989335443543203e-05, | |
| "loss": 0.3168, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1620718240737915, | |
| "step": 785, | |
| "valid_targets_mean": 6384.6, | |
| "valid_targets_min": 2678 | |
| }, | |
| { | |
| "epoch": 0.9148812970469021, | |
| "grad_norm": 0.5888450189307368, | |
| "learning_rate": 3.988731905126512e-05, | |
| "loss": 0.328, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15674595534801483, | |
| "step": 790, | |
| "valid_targets_mean": 3543.2, | |
| "valid_targets_min": 2826 | |
| }, | |
| { | |
| "epoch": 0.9206716850028952, | |
| "grad_norm": 0.7469580751777335, | |
| "learning_rate": 3.988111803708932e-05, | |
| "loss": 0.3139, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13202321529388428, | |
| "step": 795, | |
| "valid_targets_mean": 3155.8, | |
| "valid_targets_min": 993 | |
| }, | |
| { | |
| "epoch": 0.9264620729588883, | |
| "grad_norm": 0.7150531573137404, | |
| "learning_rate": 3.98747514445493e-05, | |
| "loss": 0.3218, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17962214350700378, | |
| "step": 800, | |
| "valid_targets_mean": 4166.5, | |
| "valid_targets_min": 2724 | |
| }, | |
| { | |
| "epoch": 0.9322524609148813, | |
| "grad_norm": 0.6615542925583409, | |
| "learning_rate": 3.9868219326668736e-05, | |
| "loss": 0.3224, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1518193483352661, | |
| "step": 805, | |
| "valid_targets_mean": 3079.5, | |
| "valid_targets_min": 1360 | |
| }, | |
| { | |
| "epoch": 0.9380428488708743, | |
| "grad_norm": 0.5680095143500292, | |
| "learning_rate": 3.9861521737849874e-05, | |
| "loss": 0.3163, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15517696738243103, | |
| "step": 810, | |
| "valid_targets_mean": 4114.9, | |
| "valid_targets_min": 2414 | |
| }, | |
| { | |
| "epoch": 0.9438332368268674, | |
| "grad_norm": 0.6414002936421879, | |
| "learning_rate": 3.985465873387307e-05, | |
| "loss": 0.3425, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18613509833812714, | |
| "step": 815, | |
| "valid_targets_mean": 4524.8, | |
| "valid_targets_min": 2425 | |
| }, | |
| { | |
| "epoch": 0.9496236247828604, | |
| "grad_norm": 0.7074773784167145, | |
| "learning_rate": 3.984763037189631e-05, | |
| "loss": 0.3366, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16936403512954712, | |
| "step": 820, | |
| "valid_targets_mean": 3035.2, | |
| "valid_targets_min": 1720 | |
| }, | |
| { | |
| "epoch": 0.9554140127388535, | |
| "grad_norm": 0.6703087910116017, | |
| "learning_rate": 3.984043671045479e-05, | |
| "loss": 0.3301, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1667347550392151, | |
| "step": 825, | |
| "valid_targets_mean": 3260.4, | |
| "valid_targets_min": 1621 | |
| }, | |
| { | |
| "epoch": 0.9612044006948466, | |
| "grad_norm": 0.7039481741642971, | |
| "learning_rate": 3.983307780946034e-05, | |
| "loss": 0.3121, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16874805092811584, | |
| "step": 830, | |
| "valid_targets_mean": 3214.6, | |
| "valid_targets_min": 1858 | |
| }, | |
| { | |
| "epoch": 0.9669947886508397, | |
| "grad_norm": 0.6768188627921073, | |
| "learning_rate": 3.982555373020103e-05, | |
| "loss": 0.3291, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15116912126541138, | |
| "step": 835, | |
| "valid_targets_mean": 3254.2, | |
| "valid_targets_min": 1254 | |
| }, | |
| { | |
| "epoch": 0.9727851766068326, | |
| "grad_norm": 0.6297084534327685, | |
| "learning_rate": 3.9817864535340556e-05, | |
| "loss": 0.3185, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1747068166732788, | |
| "step": 840, | |
| "valid_targets_mean": 2994.4, | |
| "valid_targets_min": 1039 | |
| }, | |
| { | |
| "epoch": 0.9785755645628257, | |
| "grad_norm": 0.6093952628926999, | |
| "learning_rate": 3.981001028891779e-05, | |
| "loss": 0.3368, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20562121272087097, | |
| "step": 845, | |
| "valid_targets_mean": 4625.4, | |
| "valid_targets_min": 2526 | |
| }, | |
| { | |
| "epoch": 0.9843659525188188, | |
| "grad_norm": 0.5915248441168647, | |
| "learning_rate": 3.980199105634623e-05, | |
| "loss": 0.3233, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17518404126167297, | |
| "step": 850, | |
| "valid_targets_mean": 3571.8, | |
| "valid_targets_min": 2409 | |
| }, | |
| { | |
| "epoch": 0.9901563404748118, | |
| "grad_norm": 0.611861719637808, | |
| "learning_rate": 3.979380690441343e-05, | |
| "loss": 0.3276, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14133447408676147, | |
| "step": 855, | |
| "valid_targets_mean": 3691.5, | |
| "valid_targets_min": 2918 | |
| }, | |
| { | |
| "epoch": 0.9959467284308049, | |
| "grad_norm": 0.6276245035498778, | |
| "learning_rate": 3.978545790128047e-05, | |
| "loss": 0.3315, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16311562061309814, | |
| "step": 860, | |
| "valid_targets_mean": 4063.1, | |
| "valid_targets_min": 2179 | |
| }, | |
| { | |
| "epoch": 1.0011580775911986, | |
| "grad_norm": 0.5249869502014928, | |
| "learning_rate": 3.9776944116481385e-05, | |
| "loss": 0.305, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13150295615196228, | |
| "step": 865, | |
| "valid_targets_mean": 8215.2, | |
| "valid_targets_min": 6163 | |
| }, | |
| { | |
| "epoch": 1.0069484655471916, | |
| "grad_norm": 0.484117327524638, | |
| "learning_rate": 3.976826562092257e-05, | |
| "loss": 0.2709, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13699647784233093, | |
| "step": 870, | |
| "valid_targets_mean": 8594.4, | |
| "valid_targets_min": 6006 | |
| }, | |
| { | |
| "epoch": 1.0127388535031847, | |
| "grad_norm": 0.5948819271463993, | |
| "learning_rate": 3.975942248688222e-05, | |
| "loss": 0.2679, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.120064377784729, | |
| "step": 875, | |
| "valid_targets_mean": 3578.4, | |
| "valid_targets_min": 289 | |
| }, | |
| { | |
| "epoch": 1.0185292414591778, | |
| "grad_norm": 0.45255503571848066, | |
| "learning_rate": 3.975041478800969e-05, | |
| "loss": 0.2582, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12774991989135742, | |
| "step": 880, | |
| "valid_targets_mean": 7019.9, | |
| "valid_targets_min": 4985 | |
| }, | |
| { | |
| "epoch": 1.0243196294151709, | |
| "grad_norm": 0.4344176759331777, | |
| "learning_rate": 3.9741242599324904e-05, | |
| "loss": 0.2381, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12448737025260925, | |
| "step": 885, | |
| "valid_targets_mean": 6956.6, | |
| "valid_targets_min": 1457 | |
| }, | |
| { | |
| "epoch": 1.030110017371164, | |
| "grad_norm": 0.4255312922643231, | |
| "learning_rate": 3.973190599721775e-05, | |
| "loss": 0.2454, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11625795066356659, | |
| "step": 890, | |
| "valid_targets_mean": 7361.6, | |
| "valid_targets_min": 4961 | |
| }, | |
| { | |
| "epoch": 1.035900405327157, | |
| "grad_norm": 0.4986133150702989, | |
| "learning_rate": 3.972240505944737e-05, | |
| "loss": 0.2627, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1547217071056366, | |
| "step": 895, | |
| "valid_targets_mean": 7058.5, | |
| "valid_targets_min": 5985 | |
| }, | |
| { | |
| "epoch": 1.04169079328315, | |
| "grad_norm": 0.4433867493523557, | |
| "learning_rate": 3.971273986514162e-05, | |
| "loss": 0.2586, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13994178175926208, | |
| "step": 900, | |
| "valid_targets_mean": 7664.2, | |
| "valid_targets_min": 5169 | |
| }, | |
| { | |
| "epoch": 1.0474811812391431, | |
| "grad_norm": 0.4274109149772053, | |
| "learning_rate": 3.9702910494796325e-05, | |
| "loss": 0.2458, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12422806024551392, | |
| "step": 905, | |
| "valid_targets_mean": 7111.8, | |
| "valid_targets_min": 5348 | |
| }, | |
| { | |
| "epoch": 1.053271569195136, | |
| "grad_norm": 0.5117732983445061, | |
| "learning_rate": 3.969291703027463e-05, | |
| "loss": 0.2441, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11535590887069702, | |
| "step": 910, | |
| "valid_targets_mean": 4238.9, | |
| "valid_targets_min": 152 | |
| }, | |
| { | |
| "epoch": 1.059061957151129, | |
| "grad_norm": 0.4501238789684571, | |
| "learning_rate": 3.968275955480635e-05, | |
| "loss": 0.2386, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12611274421215057, | |
| "step": 915, | |
| "valid_targets_mean": 7199.4, | |
| "valid_targets_min": 4258 | |
| }, | |
| { | |
| "epoch": 1.0648523451071221, | |
| "grad_norm": 0.4435191699594279, | |
| "learning_rate": 3.967243815298723e-05, | |
| "loss": 0.2412, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11774446070194244, | |
| "step": 920, | |
| "valid_targets_mean": 7243.4, | |
| "valid_targets_min": 5831 | |
| }, | |
| { | |
| "epoch": 1.0706427330631152, | |
| "grad_norm": 0.46392473348758684, | |
| "learning_rate": 3.966195291077827e-05, | |
| "loss": 0.2377, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11789186298847198, | |
| "step": 925, | |
| "valid_targets_mean": 6950.6, | |
| "valid_targets_min": 3862 | |
| }, | |
| { | |
| "epoch": 1.0764331210191083, | |
| "grad_norm": 0.4233082561923554, | |
| "learning_rate": 3.965130391550501e-05, | |
| "loss": 0.2351, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12646175920963287, | |
| "step": 930, | |
| "valid_targets_mean": 6991.9, | |
| "valid_targets_min": 5696 | |
| }, | |
| { | |
| "epoch": 1.0822235089751013, | |
| "grad_norm": 0.4367964173531144, | |
| "learning_rate": 3.9640491255856794e-05, | |
| "loss": 0.2285, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12679660320281982, | |
| "step": 935, | |
| "valid_targets_mean": 6963.5, | |
| "valid_targets_min": 4947 | |
| }, | |
| { | |
| "epoch": 1.0880138969310944, | |
| "grad_norm": 0.4320528061136158, | |
| "learning_rate": 3.962951502188601e-05, | |
| "loss": 0.2332, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11090008914470673, | |
| "step": 940, | |
| "valid_targets_mean": 6436.9, | |
| "valid_targets_min": 4988 | |
| }, | |
| { | |
| "epoch": 1.0938042848870875, | |
| "grad_norm": 0.42536386526893744, | |
| "learning_rate": 3.9618375305007394e-05, | |
| "loss": 0.2362, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1170877069234848, | |
| "step": 945, | |
| "valid_targets_mean": 7518.0, | |
| "valid_targets_min": 5476 | |
| }, | |
| { | |
| "epoch": 1.0995946728430805, | |
| "grad_norm": 0.4582340812491947, | |
| "learning_rate": 3.960707219799722e-05, | |
| "loss": 0.2279, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10813379287719727, | |
| "step": 950, | |
| "valid_targets_mean": 7422.6, | |
| "valid_targets_min": 5226 | |
| }, | |
| { | |
| "epoch": 1.1053850607990736, | |
| "grad_norm": 0.4671861431249364, | |
| "learning_rate": 3.959560579499254e-05, | |
| "loss": 0.248, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15178634226322174, | |
| "step": 955, | |
| "valid_targets_mean": 7080.0, | |
| "valid_targets_min": 3753 | |
| }, | |
| { | |
| "epoch": 1.1111754487550667, | |
| "grad_norm": 0.5169416581408465, | |
| "learning_rate": 3.9583976191490426e-05, | |
| "loss": 0.2594, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13366922736167908, | |
| "step": 960, | |
| "valid_targets_mean": 6064.6, | |
| "valid_targets_min": 5161 | |
| }, | |
| { | |
| "epoch": 1.1169658367110595, | |
| "grad_norm": 0.418553522605153, | |
| "learning_rate": 3.957218348434711e-05, | |
| "loss": 0.2175, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11265334486961365, | |
| "step": 965, | |
| "valid_targets_mean": 7388.1, | |
| "valid_targets_min": 5795 | |
| }, | |
| { | |
| "epoch": 1.1227562246670526, | |
| "grad_norm": 0.388992424482383, | |
| "learning_rate": 3.956022777177727e-05, | |
| "loss": 0.2221, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10571281611919403, | |
| "step": 970, | |
| "valid_targets_mean": 7696.0, | |
| "valid_targets_min": 4658 | |
| }, | |
| { | |
| "epoch": 1.1285466126230457, | |
| "grad_norm": 0.4681809942636834, | |
| "learning_rate": 3.954810915335314e-05, | |
| "loss": 0.2286, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1254599690437317, | |
| "step": 975, | |
| "valid_targets_mean": 7799.1, | |
| "valid_targets_min": 5508 | |
| }, | |
| { | |
| "epoch": 1.1343370005790387, | |
| "grad_norm": 0.48398865515756395, | |
| "learning_rate": 3.9535827730003686e-05, | |
| "loss": 0.2401, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11612563580274582, | |
| "step": 980, | |
| "valid_targets_mean": 7341.6, | |
| "valid_targets_min": 6454 | |
| }, | |
| { | |
| "epoch": 1.1401273885350318, | |
| "grad_norm": 0.4490116781150715, | |
| "learning_rate": 3.952338360401382e-05, | |
| "loss": 0.2308, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10329363495111465, | |
| "step": 985, | |
| "valid_targets_mean": 6818.0, | |
| "valid_targets_min": 4392 | |
| }, | |
| { | |
| "epoch": 1.1459177764910249, | |
| "grad_norm": 0.41953032426909853, | |
| "learning_rate": 3.951077687902347e-05, | |
| "loss": 0.2299, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10158362239599228, | |
| "step": 990, | |
| "valid_targets_mean": 7857.6, | |
| "valid_targets_min": 5394 | |
| }, | |
| { | |
| "epoch": 1.151708164447018, | |
| "grad_norm": 0.3947288696909538, | |
| "learning_rate": 3.949800766002679e-05, | |
| "loss": 0.2013, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09270112216472626, | |
| "step": 995, | |
| "valid_targets_mean": 8314.1, | |
| "valid_targets_min": 4837 | |
| }, | |
| { | |
| "epoch": 1.157498552403011, | |
| "grad_norm": 0.36433462079153023, | |
| "learning_rate": 3.9485076053371234e-05, | |
| "loss": 0.2036, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08710625767707825, | |
| "step": 1000, | |
| "valid_targets_mean": 7368.2, | |
| "valid_targets_min": 6027 | |
| }, | |
| { | |
| "epoch": 1.163288940359004, | |
| "grad_norm": 0.45419044499402866, | |
| "learning_rate": 3.9471982166756686e-05, | |
| "loss": 0.2132, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11508241295814514, | |
| "step": 1005, | |
| "valid_targets_mean": 6435.0, | |
| "valid_targets_min": 4238 | |
| }, | |
| { | |
| "epoch": 1.1690793283149972, | |
| "grad_norm": 2.111154060538335, | |
| "learning_rate": 3.9458726109234595e-05, | |
| "loss": 0.235, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09870345890522003, | |
| "step": 1010, | |
| "valid_targets_mean": 247.8, | |
| "valid_targets_min": 137 | |
| }, | |
| { | |
| "epoch": 1.1748697162709902, | |
| "grad_norm": 0.45137108403630927, | |
| "learning_rate": 3.9445307991206993e-05, | |
| "loss": 0.2396, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11088582128286362, | |
| "step": 1015, | |
| "valid_targets_mean": 6807.0, | |
| "valid_targets_min": 5165 | |
| }, | |
| { | |
| "epoch": 1.180660104226983, | |
| "grad_norm": 0.8712175642859009, | |
| "learning_rate": 3.943172792442567e-05, | |
| "loss": 0.2438, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12714217603206635, | |
| "step": 1020, | |
| "valid_targets_mean": 6989.9, | |
| "valid_targets_min": 5104 | |
| }, | |
| { | |
| "epoch": 1.1864504921829764, | |
| "grad_norm": 0.4413781971226211, | |
| "learning_rate": 3.941798602199115e-05, | |
| "loss": 0.2421, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13100311160087585, | |
| "step": 1025, | |
| "valid_targets_mean": 6692.2, | |
| "valid_targets_min": 5068 | |
| }, | |
| { | |
| "epoch": 1.1922408801389692, | |
| "grad_norm": 0.4435462481029505, | |
| "learning_rate": 3.9404082398351825e-05, | |
| "loss": 0.2355, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14060164988040924, | |
| "step": 1030, | |
| "valid_targets_mean": 7419.6, | |
| "valid_targets_min": 5578 | |
| }, | |
| { | |
| "epoch": 1.1980312680949623, | |
| "grad_norm": 0.4060121325738929, | |
| "learning_rate": 3.939001716930294e-05, | |
| "loss": 0.2249, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11470702290534973, | |
| "step": 1035, | |
| "valid_targets_mean": 7536.5, | |
| "valid_targets_min": 5075 | |
| }, | |
| { | |
| "epoch": 1.2038216560509554, | |
| "grad_norm": 0.38409709927516766, | |
| "learning_rate": 3.93757904519857e-05, | |
| "loss": 0.2338, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10600236058235168, | |
| "step": 1040, | |
| "valid_targets_mean": 7221.5, | |
| "valid_targets_min": 5841 | |
| }, | |
| { | |
| "epoch": 1.2096120440069484, | |
| "grad_norm": 0.4056910276302505, | |
| "learning_rate": 3.93614023648862e-05, | |
| "loss": 0.2272, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11286841332912445, | |
| "step": 1045, | |
| "valid_targets_mean": 7287.5, | |
| "valid_targets_min": 5151 | |
| }, | |
| { | |
| "epoch": 1.2154024319629415, | |
| "grad_norm": 0.4627842508865308, | |
| "learning_rate": 3.934685302783454e-05, | |
| "loss": 0.2383, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11054033041000366, | |
| "step": 1050, | |
| "valid_targets_mean": 6738.4, | |
| "valid_targets_min": 4042 | |
| }, | |
| { | |
| "epoch": 1.2211928199189346, | |
| "grad_norm": 0.4534271435797926, | |
| "learning_rate": 3.9332142562003735e-05, | |
| "loss": 0.2284, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12786555290222168, | |
| "step": 1055, | |
| "valid_targets_mean": 6331.0, | |
| "valid_targets_min": 5496 | |
| }, | |
| { | |
| "epoch": 1.2269832078749277, | |
| "grad_norm": 0.4673448734063373, | |
| "learning_rate": 3.9317271089908784e-05, | |
| "loss": 0.2318, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1133434921503067, | |
| "step": 1060, | |
| "valid_targets_mean": 6271.1, | |
| "valid_targets_min": 4695 | |
| }, | |
| { | |
| "epoch": 1.2327735958309207, | |
| "grad_norm": 0.49865251236737174, | |
| "learning_rate": 3.93022387354056e-05, | |
| "loss": 0.2325, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12024211138486862, | |
| "step": 1065, | |
| "valid_targets_mean": 5790.8, | |
| "valid_targets_min": 5021 | |
| }, | |
| { | |
| "epoch": 1.2385639837869138, | |
| "grad_norm": 0.5926834734740676, | |
| "learning_rate": 3.9287045623689985e-05, | |
| "loss": 0.2237, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09189209342002869, | |
| "step": 1070, | |
| "valid_targets_mean": 3386.8, | |
| "valid_targets_min": 2279 | |
| }, | |
| { | |
| "epoch": 1.2443543717429069, | |
| "grad_norm": 0.4794203925484133, | |
| "learning_rate": 3.927169188129661e-05, | |
| "loss": 0.2234, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11206869035959244, | |
| "step": 1075, | |
| "valid_targets_mean": 6240.5, | |
| "valid_targets_min": 5093 | |
| }, | |
| { | |
| "epoch": 1.2501447596989, | |
| "grad_norm": 0.4632615561507731, | |
| "learning_rate": 3.9256177636097934e-05, | |
| "loss": 0.2255, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11752070486545563, | |
| "step": 1080, | |
| "valid_targets_mean": 6329.5, | |
| "valid_targets_min": 4589 | |
| }, | |
| { | |
| "epoch": 1.2559351476548928, | |
| "grad_norm": 2.0458169274072926, | |
| "learning_rate": 3.924050301730316e-05, | |
| "loss": 0.2241, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12004910409450531, | |
| "step": 1085, | |
| "valid_targets_mean": 7589.2, | |
| "valid_targets_min": 4963 | |
| }, | |
| { | |
| "epoch": 1.2617255356108859, | |
| "grad_norm": 0.42400886081444505, | |
| "learning_rate": 3.9224668155457146e-05, | |
| "loss": 0.2251, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09688187390565872, | |
| "step": 1090, | |
| "valid_targets_mean": 5577.9, | |
| "valid_targets_min": 4458 | |
| }, | |
| { | |
| "epoch": 1.267515923566879, | |
| "grad_norm": 0.5540750132695128, | |
| "learning_rate": 3.920867318243931e-05, | |
| "loss": 0.2237, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10729500651359558, | |
| "step": 1095, | |
| "valid_targets_mean": 5432.5, | |
| "valid_targets_min": 4433 | |
| }, | |
| { | |
| "epoch": 1.273306311522872, | |
| "grad_norm": 0.44705611186587985, | |
| "learning_rate": 3.9192518231462585e-05, | |
| "loss": 0.2277, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11662057787179947, | |
| "step": 1100, | |
| "valid_targets_mean": 6361.8, | |
| "valid_targets_min": 5201 | |
| }, | |
| { | |
| "epoch": 1.279096699478865, | |
| "grad_norm": 0.4549497220381097, | |
| "learning_rate": 3.917620343707221e-05, | |
| "loss": 0.2285, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1052645817399025, | |
| "step": 1105, | |
| "valid_targets_mean": 5824.4, | |
| "valid_targets_min": 5129 | |
| }, | |
| { | |
| "epoch": 1.2848870874348581, | |
| "grad_norm": 0.9022835962871019, | |
| "learning_rate": 3.915972893514471e-05, | |
| "loss": 0.2018, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1566964089870453, | |
| "step": 1110, | |
| "valid_targets_mean": 1637.6, | |
| "valid_targets_min": 137 | |
| }, | |
| { | |
| "epoch": 1.2906774753908512, | |
| "grad_norm": 0.46381011488635515, | |
| "learning_rate": 3.9143094862886705e-05, | |
| "loss": 0.2241, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11298874765634537, | |
| "step": 1115, | |
| "valid_targets_mean": 6474.6, | |
| "valid_targets_min": 4751 | |
| }, | |
| { | |
| "epoch": 1.2964678633468443, | |
| "grad_norm": 0.4472651737046458, | |
| "learning_rate": 3.9126301358833786e-05, | |
| "loss": 0.2205, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12175101041793823, | |
| "step": 1120, | |
| "valid_targets_mean": 7186.2, | |
| "valid_targets_min": 5358 | |
| }, | |
| { | |
| "epoch": 1.3022582513028373, | |
| "grad_norm": 0.4552798202628588, | |
| "learning_rate": 3.910934856284937e-05, | |
| "loss": 0.2268, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11371338367462158, | |
| "step": 1125, | |
| "valid_targets_mean": 7126.8, | |
| "valid_targets_min": 4078 | |
| }, | |
| { | |
| "epoch": 1.3080486392588304, | |
| "grad_norm": 0.4564388018494831, | |
| "learning_rate": 3.90922366161235e-05, | |
| "loss": 0.2183, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10979453474283218, | |
| "step": 1130, | |
| "valid_targets_mean": 7022.1, | |
| "valid_targets_min": 4940 | |
| }, | |
| { | |
| "epoch": 1.3138390272148235, | |
| "grad_norm": 0.44457997970128343, | |
| "learning_rate": 3.907496566117173e-05, | |
| "loss": 0.2133, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11380468308925629, | |
| "step": 1135, | |
| "valid_targets_mean": 5900.6, | |
| "valid_targets_min": 4779 | |
| }, | |
| { | |
| "epoch": 1.3196294151708163, | |
| "grad_norm": 0.44167393228026747, | |
| "learning_rate": 3.905753584183387e-05, | |
| "loss": 0.2233, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11499583721160889, | |
| "step": 1140, | |
| "valid_targets_mean": 6290.1, | |
| "valid_targets_min": 4620 | |
| }, | |
| { | |
| "epoch": 1.3254198031268096, | |
| "grad_norm": 0.5428718455705244, | |
| "learning_rate": 3.9039947303272836e-05, | |
| "loss": 0.2876, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1602226346731186, | |
| "step": 1145, | |
| "valid_targets_mean": 5614.5, | |
| "valid_targets_min": 3446 | |
| }, | |
| { | |
| "epoch": 1.3312101910828025, | |
| "grad_norm": 0.6699882200896025, | |
| "learning_rate": 3.902220019197342e-05, | |
| "loss": 0.3224, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16955219209194183, | |
| "step": 1150, | |
| "valid_targets_mean": 4630.0, | |
| "valid_targets_min": 1005 | |
| }, | |
| { | |
| "epoch": 1.3370005790387955, | |
| "grad_norm": 0.5145366457651853, | |
| "learning_rate": 3.900429465574106e-05, | |
| "loss": 0.3115, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11790046095848083, | |
| "step": 1155, | |
| "valid_targets_mean": 4669.6, | |
| "valid_targets_min": 2054 | |
| }, | |
| { | |
| "epoch": 1.3427909669947886, | |
| "grad_norm": 0.5699015253947644, | |
| "learning_rate": 3.898623084370066e-05, | |
| "loss": 0.3097, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1340475082397461, | |
| "step": 1160, | |
| "valid_targets_mean": 3745.0, | |
| "valid_targets_min": 1561 | |
| }, | |
| { | |
| "epoch": 1.3485813549507817, | |
| "grad_norm": 0.5755210822032388, | |
| "learning_rate": 3.8968008906295274e-05, | |
| "loss": 0.3064, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1492936909198761, | |
| "step": 1165, | |
| "valid_targets_mean": 4284.0, | |
| "valid_targets_min": 1184 | |
| }, | |
| { | |
| "epoch": 1.3543717429067748, | |
| "grad_norm": 0.4916120473926803, | |
| "learning_rate": 3.894962899528492e-05, | |
| "loss": 0.3211, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16478881239891052, | |
| "step": 1170, | |
| "valid_targets_mean": 6332.8, | |
| "valid_targets_min": 1651 | |
| }, | |
| { | |
| "epoch": 1.3601621308627678, | |
| "grad_norm": 0.46790533699331943, | |
| "learning_rate": 3.893109126374528e-05, | |
| "loss": 0.3042, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14911609888076782, | |
| "step": 1175, | |
| "valid_targets_mean": 6446.5, | |
| "valid_targets_min": 2231 | |
| }, | |
| { | |
| "epoch": 1.365952518818761, | |
| "grad_norm": 0.5381056441690047, | |
| "learning_rate": 3.8912395866066404e-05, | |
| "loss": 0.3065, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11903347074985504, | |
| "step": 1180, | |
| "valid_targets_mean": 3367.2, | |
| "valid_targets_min": 467 | |
| }, | |
| { | |
| "epoch": 1.371742906774754, | |
| "grad_norm": 0.6066646913438828, | |
| "learning_rate": 3.8893542957951496e-05, | |
| "loss": 0.3107, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1923138052225113, | |
| "step": 1185, | |
| "valid_targets_mean": 7194.1, | |
| "valid_targets_min": 1748 | |
| }, | |
| { | |
| "epoch": 1.377533294730747, | |
| "grad_norm": 0.5901290970032714, | |
| "learning_rate": 3.8874532696415544e-05, | |
| "loss": 0.3193, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16743800044059753, | |
| "step": 1190, | |
| "valid_targets_mean": 4078.5, | |
| "valid_targets_min": 1697 | |
| }, | |
| { | |
| "epoch": 1.3833236826867399, | |
| "grad_norm": 0.7113483233393854, | |
| "learning_rate": 3.8855365239784055e-05, | |
| "loss": 0.3117, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1085222065448761, | |
| "step": 1195, | |
| "valid_targets_mean": 2089.5, | |
| "valid_targets_min": 1223 | |
| }, | |
| { | |
| "epoch": 1.3891140706427332, | |
| "grad_norm": 0.49830543569902624, | |
| "learning_rate": 3.883604074769172e-05, | |
| "loss": 0.3064, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1353009045124054, | |
| "step": 1200, | |
| "valid_targets_mean": 5404.1, | |
| "valid_targets_min": 1829 | |
| }, | |
| { | |
| "epoch": 1.394904458598726, | |
| "grad_norm": 0.49106905409625246, | |
| "learning_rate": 3.8816559381081077e-05, | |
| "loss": 0.3128, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1435467153787613, | |
| "step": 1205, | |
| "valid_targets_mean": 4583.1, | |
| "valid_targets_min": 1596 | |
| }, | |
| { | |
| "epoch": 1.400694846554719, | |
| "grad_norm": 0.599595031027154, | |
| "learning_rate": 3.8796921302201195e-05, | |
| "loss": 0.3076, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17589807510375977, | |
| "step": 1210, | |
| "valid_targets_mean": 4666.8, | |
| "valid_targets_min": 2041 | |
| }, | |
| { | |
| "epoch": 1.4064852345107122, | |
| "grad_norm": 0.6023402305027469, | |
| "learning_rate": 3.877712667460631e-05, | |
| "loss": 0.321, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17788101732730865, | |
| "step": 1215, | |
| "valid_targets_mean": 4272.1, | |
| "valid_targets_min": 1538 | |
| }, | |
| { | |
| "epoch": 1.4122756224667052, | |
| "grad_norm": 0.6410743256350347, | |
| "learning_rate": 3.875717566315446e-05, | |
| "loss": 0.3152, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14403772354125977, | |
| "step": 1220, | |
| "valid_targets_mean": 2928.0, | |
| "valid_targets_min": 1739 | |
| }, | |
| { | |
| "epoch": 1.4180660104226983, | |
| "grad_norm": 0.6508045068605425, | |
| "learning_rate": 3.87370684340061e-05, | |
| "loss": 0.3113, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14435258507728577, | |
| "step": 1225, | |
| "valid_targets_mean": 3254.5, | |
| "valid_targets_min": 935 | |
| }, | |
| { | |
| "epoch": 1.4238563983786914, | |
| "grad_norm": 0.6830096233724313, | |
| "learning_rate": 3.8716805154622756e-05, | |
| "loss": 0.3236, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15719226002693176, | |
| "step": 1230, | |
| "valid_targets_mean": 3535.2, | |
| "valid_targets_min": 2704 | |
| }, | |
| { | |
| "epoch": 1.4296467863346844, | |
| "grad_norm": 0.7793111606201882, | |
| "learning_rate": 3.869638599376558e-05, | |
| "loss": 0.3215, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.156528502702713, | |
| "step": 1235, | |
| "valid_targets_mean": 2987.6, | |
| "valid_targets_min": 1173 | |
| }, | |
| { | |
| "epoch": 1.4354371742906775, | |
| "grad_norm": 0.5927550164432058, | |
| "learning_rate": 3.8675811121494e-05, | |
| "loss": 0.3186, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14001598954200745, | |
| "step": 1240, | |
| "valid_targets_mean": 3700.6, | |
| "valid_targets_min": 1670 | |
| }, | |
| { | |
| "epoch": 1.4412275622466706, | |
| "grad_norm": 0.6055360826788108, | |
| "learning_rate": 3.865508070916424e-05, | |
| "loss": 0.3186, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14109604060649872, | |
| "step": 1245, | |
| "valid_targets_mean": 3481.6, | |
| "valid_targets_min": 1329 | |
| }, | |
| { | |
| "epoch": 1.4470179502026637, | |
| "grad_norm": 0.629972581855596, | |
| "learning_rate": 3.863419492942797e-05, | |
| "loss": 0.3075, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16355670988559723, | |
| "step": 1250, | |
| "valid_targets_mean": 3589.8, | |
| "valid_targets_min": 2279 | |
| }, | |
| { | |
| "epoch": 1.4528083381586567, | |
| "grad_norm": 0.5165126632126333, | |
| "learning_rate": 3.861315395623078e-05, | |
| "loss": 0.2881, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1432054042816162, | |
| "step": 1255, | |
| "valid_targets_mean": 5060.4, | |
| "valid_targets_min": 2222 | |
| }, | |
| { | |
| "epoch": 1.4585987261146496, | |
| "grad_norm": 0.5781860344801439, | |
| "learning_rate": 3.85919579648108e-05, | |
| "loss": 0.3132, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15431106090545654, | |
| "step": 1260, | |
| "valid_targets_mean": 3898.6, | |
| "valid_targets_min": 1849 | |
| }, | |
| { | |
| "epoch": 1.4643891140706427, | |
| "grad_norm": 0.5956044360840919, | |
| "learning_rate": 3.857060713169724e-05, | |
| "loss": 0.3117, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13421255350112915, | |
| "step": 1265, | |
| "valid_targets_mean": 3698.9, | |
| "valid_targets_min": 1951 | |
| }, | |
| { | |
| "epoch": 1.4701795020266357, | |
| "grad_norm": 0.5546382516771821, | |
| "learning_rate": 3.854910163470885e-05, | |
| "loss": 0.3201, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16746890544891357, | |
| "step": 1270, | |
| "valid_targets_mean": 4506.8, | |
| "valid_targets_min": 1765 | |
| }, | |
| { | |
| "epoch": 1.4759698899826288, | |
| "grad_norm": 0.6775028329226226, | |
| "learning_rate": 3.852744165295252e-05, | |
| "loss": 0.2982, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14411398768424988, | |
| "step": 1275, | |
| "valid_targets_mean": 2991.6, | |
| "valid_targets_min": 2087 | |
| }, | |
| { | |
| "epoch": 1.4817602779386219, | |
| "grad_norm": 0.561251134139609, | |
| "learning_rate": 3.850562736682173e-05, | |
| "loss": 0.2988, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13625818490982056, | |
| "step": 1280, | |
| "valid_targets_mean": 3994.1, | |
| "valid_targets_min": 1603 | |
| }, | |
| { | |
| "epoch": 1.487550665894615, | |
| "grad_norm": 0.6083852437758633, | |
| "learning_rate": 3.8483658957995114e-05, | |
| "loss": 0.311, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16182228922843933, | |
| "step": 1285, | |
| "valid_targets_mean": 4391.9, | |
| "valid_targets_min": 2092 | |
| }, | |
| { | |
| "epoch": 1.493341053850608, | |
| "grad_norm": 0.5913661335465749, | |
| "learning_rate": 3.8461536609434855e-05, | |
| "loss": 0.3043, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15757182240486145, | |
| "step": 1290, | |
| "valid_targets_mean": 3808.9, | |
| "valid_targets_min": 1410 | |
| }, | |
| { | |
| "epoch": 1.499131441806601, | |
| "grad_norm": 0.5394656768823203, | |
| "learning_rate": 3.843926050538524e-05, | |
| "loss": 0.2974, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16474880278110504, | |
| "step": 1295, | |
| "valid_targets_mean": 5078.5, | |
| "valid_targets_min": 2110 | |
| }, | |
| { | |
| "epoch": 1.5049218297625941, | |
| "grad_norm": 0.6243755838275791, | |
| "learning_rate": 3.841683083137108e-05, | |
| "loss": 0.3068, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14915981888771057, | |
| "step": 1300, | |
| "valid_targets_mean": 3447.4, | |
| "valid_targets_min": 2420 | |
| }, | |
| { | |
| "epoch": 1.510712217718587, | |
| "grad_norm": 0.6375167647764206, | |
| "learning_rate": 3.8394247774196206e-05, | |
| "loss": 0.306, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14488837122917175, | |
| "step": 1305, | |
| "valid_targets_mean": 4371.1, | |
| "valid_targets_min": 1876 | |
| }, | |
| { | |
| "epoch": 1.5165026056745803, | |
| "grad_norm": 0.613454201006315, | |
| "learning_rate": 3.837151152194187e-05, | |
| "loss": 0.3018, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17439305782318115, | |
| "step": 1310, | |
| "valid_targets_mean": 3858.2, | |
| "valid_targets_min": 2605 | |
| }, | |
| { | |
| "epoch": 1.5222929936305731, | |
| "grad_norm": 0.6233105779650496, | |
| "learning_rate": 3.83486222639652e-05, | |
| "loss": 0.311, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13862144947052002, | |
| "step": 1315, | |
| "valid_targets_mean": 3383.9, | |
| "valid_targets_min": 1177 | |
| }, | |
| { | |
| "epoch": 1.5280833815865664, | |
| "grad_norm": 0.6467891702920507, | |
| "learning_rate": 3.832558019089763e-05, | |
| "loss": 0.3099, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15471479296684265, | |
| "step": 1320, | |
| "valid_targets_mean": 3455.5, | |
| "valid_targets_min": 1714 | |
| }, | |
| { | |
| "epoch": 1.5338737695425593, | |
| "grad_norm": 0.5289820990421887, | |
| "learning_rate": 3.830238549464329e-05, | |
| "loss": 0.3015, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16986386477947235, | |
| "step": 1325, | |
| "valid_targets_mean": 5547.5, | |
| "valid_targets_min": 2033 | |
| }, | |
| { | |
| "epoch": 1.5396641574985526, | |
| "grad_norm": 0.7060162824223625, | |
| "learning_rate": 3.8279038368377415e-05, | |
| "loss": 0.297, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.161880761384964, | |
| "step": 1330, | |
| "valid_targets_mean": 3485.4, | |
| "valid_targets_min": 1924 | |
| }, | |
| { | |
| "epoch": 1.5454545454545454, | |
| "grad_norm": 0.68455855185045, | |
| "learning_rate": 3.825553900654478e-05, | |
| "loss": 0.3018, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13728755712509155, | |
| "step": 1335, | |
| "valid_targets_mean": 3080.2, | |
| "valid_targets_min": 2098 | |
| }, | |
| { | |
| "epoch": 1.5512449334105385, | |
| "grad_norm": 0.7207851738392093, | |
| "learning_rate": 3.823188760485798e-05, | |
| "loss": 0.3025, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17264071106910706, | |
| "step": 1340, | |
| "valid_targets_mean": 3393.4, | |
| "valid_targets_min": 1363 | |
| }, | |
| { | |
| "epoch": 1.5570353213665316, | |
| "grad_norm": 0.6036114150424443, | |
| "learning_rate": 3.820808436029593e-05, | |
| "loss": 0.302, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16787129640579224, | |
| "step": 1345, | |
| "valid_targets_mean": 4115.4, | |
| "valid_targets_min": 1363 | |
| }, | |
| { | |
| "epoch": 1.5628257093225246, | |
| "grad_norm": 0.595651877586518, | |
| "learning_rate": 3.81841294711021e-05, | |
| "loss": 0.306, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17603866755962372, | |
| "step": 1350, | |
| "valid_targets_mean": 4313.8, | |
| "valid_targets_min": 2620 | |
| }, | |
| { | |
| "epoch": 1.5686160972785177, | |
| "grad_norm": 0.6336909981450094, | |
| "learning_rate": 3.8160023136782956e-05, | |
| "loss": 0.3011, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15865519642829895, | |
| "step": 1355, | |
| "valid_targets_mean": 3425.1, | |
| "valid_targets_min": 2316 | |
| }, | |
| { | |
| "epoch": 1.5744064852345108, | |
| "grad_norm": 0.6641668751975259, | |
| "learning_rate": 3.813576555810625e-05, | |
| "loss": 0.2965, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13947749137878418, | |
| "step": 1360, | |
| "valid_targets_mean": 3744.0, | |
| "valid_targets_min": 2022 | |
| }, | |
| { | |
| "epoch": 1.5801968731905038, | |
| "grad_norm": 0.6430114196277706, | |
| "learning_rate": 3.811135693709935e-05, | |
| "loss": 0.2891, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10136308521032333, | |
| "step": 1365, | |
| "valid_targets_mean": 2445.6, | |
| "valid_targets_min": 1294 | |
| }, | |
| { | |
| "epoch": 1.5859872611464967, | |
| "grad_norm": 0.611201336181961, | |
| "learning_rate": 3.80867974770476e-05, | |
| "loss": 0.3026, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14630404114723206, | |
| "step": 1370, | |
| "valid_targets_mean": 3556.1, | |
| "valid_targets_min": 1700 | |
| }, | |
| { | |
| "epoch": 1.59177764910249, | |
| "grad_norm": 0.6115984226936316, | |
| "learning_rate": 3.806208738249257e-05, | |
| "loss": 0.2945, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1197759360074997, | |
| "step": 1375, | |
| "valid_targets_mean": 2714.0, | |
| "valid_targets_min": 1586 | |
| }, | |
| { | |
| "epoch": 1.5975680370584828, | |
| "grad_norm": 0.689916065021143, | |
| "learning_rate": 3.803722685923038e-05, | |
| "loss": 0.3001, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16724331676959991, | |
| "step": 1380, | |
| "valid_targets_mean": 3266.4, | |
| "valid_targets_min": 1358 | |
| }, | |
| { | |
| "epoch": 1.6033584250144761, | |
| "grad_norm": 0.6326874996965296, | |
| "learning_rate": 3.801221611430999e-05, | |
| "loss": 0.3034, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14491121470928192, | |
| "step": 1385, | |
| "valid_targets_mean": 3782.9, | |
| "valid_targets_min": 2290 | |
| }, | |
| { | |
| "epoch": 1.609148812970469, | |
| "grad_norm": 0.636924161233398, | |
| "learning_rate": 3.798705535603146e-05, | |
| "loss": 0.305, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19842056930065155, | |
| "step": 1390, | |
| "valid_targets_mean": 4904.8, | |
| "valid_targets_min": 2189 | |
| }, | |
| { | |
| "epoch": 1.614939200926462, | |
| "grad_norm": 0.617435396839824, | |
| "learning_rate": 3.796174479394425e-05, | |
| "loss": 0.2959, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19284765422344208, | |
| "step": 1395, | |
| "valid_targets_mean": 3959.5, | |
| "valid_targets_min": 2121 | |
| }, | |
| { | |
| "epoch": 1.620729588882455, | |
| "grad_norm": 0.6063095285796238, | |
| "learning_rate": 3.7936284638845416e-05, | |
| "loss": 0.3026, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15117163956165314, | |
| "step": 1400, | |
| "valid_targets_mean": 3689.4, | |
| "valid_targets_min": 2011 | |
| }, | |
| { | |
| "epoch": 1.6265199768384482, | |
| "grad_norm": 0.5590404864745255, | |
| "learning_rate": 3.791067510277792e-05, | |
| "loss": 0.2921, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16714683175086975, | |
| "step": 1405, | |
| "valid_targets_mean": 5082.1, | |
| "valid_targets_min": 2674 | |
| }, | |
| { | |
| "epoch": 1.6323103647944412, | |
| "grad_norm": 0.6010907769653863, | |
| "learning_rate": 3.7884916399028796e-05, | |
| "loss": 0.2824, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16495585441589355, | |
| "step": 1410, | |
| "valid_targets_mean": 3660.9, | |
| "valid_targets_min": 2266 | |
| }, | |
| { | |
| "epoch": 1.6381007527504343, | |
| "grad_norm": 0.5413580426914476, | |
| "learning_rate": 3.785900874212744e-05, | |
| "loss": 0.2868, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1378852128982544, | |
| "step": 1415, | |
| "valid_targets_mean": 4285.5, | |
| "valid_targets_min": 904 | |
| }, | |
| { | |
| "epoch": 1.6438911407064274, | |
| "grad_norm": 0.5985754205236895, | |
| "learning_rate": 3.783295234784378e-05, | |
| "loss": 0.2906, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16695967316627502, | |
| "step": 1420, | |
| "valid_targets_mean": 3999.4, | |
| "valid_targets_min": 1194 | |
| }, | |
| { | |
| "epoch": 1.6496815286624202, | |
| "grad_norm": 0.5642800803882113, | |
| "learning_rate": 3.78067474331865e-05, | |
| "loss": 0.2957, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13066710531711578, | |
| "step": 1425, | |
| "valid_targets_mean": 4011.8, | |
| "valid_targets_min": 1658 | |
| }, | |
| { | |
| "epoch": 1.6554719166184135, | |
| "grad_norm": 0.6748860615080935, | |
| "learning_rate": 3.778039421640121e-05, | |
| "loss": 0.2935, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12626111507415771, | |
| "step": 1430, | |
| "valid_targets_mean": 2728.8, | |
| "valid_targets_min": 1423 | |
| }, | |
| { | |
| "epoch": 1.6612623045744064, | |
| "grad_norm": 0.6061991472050072, | |
| "learning_rate": 3.775389291696866e-05, | |
| "loss": 0.2947, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1608344465494156, | |
| "step": 1435, | |
| "valid_targets_mean": 4504.4, | |
| "valid_targets_min": 2320 | |
| }, | |
| { | |
| "epoch": 1.6670526925303997, | |
| "grad_norm": 0.6011184485544011, | |
| "learning_rate": 3.7727243755602874e-05, | |
| "loss": 0.2985, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11193178594112396, | |
| "step": 1440, | |
| "valid_targets_mean": 3272.5, | |
| "valid_targets_min": 1811 | |
| }, | |
| { | |
| "epoch": 1.6728430804863925, | |
| "grad_norm": 0.5264073883119884, | |
| "learning_rate": 3.770044695424935e-05, | |
| "loss": 0.3039, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1393248736858368, | |
| "step": 1445, | |
| "valid_targets_mean": 4425.9, | |
| "valid_targets_min": 3026 | |
| }, | |
| { | |
| "epoch": 1.6786334684423856, | |
| "grad_norm": 0.6441290731247149, | |
| "learning_rate": 3.76735027360832e-05, | |
| "loss": 0.2948, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12856018543243408, | |
| "step": 1450, | |
| "valid_targets_mean": 2719.1, | |
| "valid_targets_min": 644 | |
| }, | |
| { | |
| "epoch": 1.6844238563983787, | |
| "grad_norm": 0.5795525157839163, | |
| "learning_rate": 3.764641132550726e-05, | |
| "loss": 0.2857, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1311633586883545, | |
| "step": 1455, | |
| "valid_targets_mean": 3369.0, | |
| "valid_targets_min": 1397 | |
| }, | |
| { | |
| "epoch": 1.6902142443543717, | |
| "grad_norm": 0.6501614001474612, | |
| "learning_rate": 3.761917294815028e-05, | |
| "loss": 0.2775, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14685682952404022, | |
| "step": 1460, | |
| "valid_targets_mean": 4122.6, | |
| "valid_targets_min": 1415 | |
| }, | |
| { | |
| "epoch": 1.6960046323103648, | |
| "grad_norm": 0.6270974962784911, | |
| "learning_rate": 3.759178783086498e-05, | |
| "loss": 0.2869, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1565871238708496, | |
| "step": 1465, | |
| "valid_targets_mean": 4136.6, | |
| "valid_targets_min": 621 | |
| }, | |
| { | |
| "epoch": 1.7017950202663579, | |
| "grad_norm": 0.6060924529762027, | |
| "learning_rate": 3.75642562017262e-05, | |
| "loss": 0.2843, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1117183268070221, | |
| "step": 1470, | |
| "valid_targets_mean": 3065.8, | |
| "valid_targets_min": 1290 | |
| }, | |
| { | |
| "epoch": 1.707585408222351, | |
| "grad_norm": 0.6638246415488404, | |
| "learning_rate": 3.7536578290029e-05, | |
| "loss": 0.2744, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12210803478956223, | |
| "step": 1475, | |
| "valid_targets_mean": 2426.9, | |
| "valid_targets_min": 857 | |
| }, | |
| { | |
| "epoch": 1.7133757961783438, | |
| "grad_norm": 0.7507078491383454, | |
| "learning_rate": 3.750875432628676e-05, | |
| "loss": 0.2789, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15053801238536835, | |
| "step": 1480, | |
| "valid_targets_mean": 2949.1, | |
| "valid_targets_min": 720 | |
| }, | |
| { | |
| "epoch": 1.719166184134337, | |
| "grad_norm": 0.625818585702717, | |
| "learning_rate": 3.74807845422292e-05, | |
| "loss": 0.2891, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16986438632011414, | |
| "step": 1485, | |
| "valid_targets_mean": 3579.1, | |
| "valid_targets_min": 1801 | |
| }, | |
| { | |
| "epoch": 1.72495657209033, | |
| "grad_norm": 0.6211865498978847, | |
| "learning_rate": 3.745266917080054e-05, | |
| "loss": 0.3079, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14206188917160034, | |
| "step": 1490, | |
| "valid_targets_mean": 3532.6, | |
| "valid_targets_min": 1510 | |
| }, | |
| { | |
| "epoch": 1.7307469600463232, | |
| "grad_norm": 0.6918628290457827, | |
| "learning_rate": 3.742440844615747e-05, | |
| "loss": 0.2831, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13704082369804382, | |
| "step": 1495, | |
| "valid_targets_mean": 2807.9, | |
| "valid_targets_min": 987 | |
| }, | |
| { | |
| "epoch": 1.736537348002316, | |
| "grad_norm": 0.6662912463540775, | |
| "learning_rate": 3.739600260366728e-05, | |
| "loss": 0.294, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12647037208080292, | |
| "step": 1500, | |
| "valid_targets_mean": 3937.4, | |
| "valid_targets_min": 996 | |
| }, | |
| { | |
| "epoch": 1.7423277359583094, | |
| "grad_norm": 0.5716296457229726, | |
| "learning_rate": 3.7367451879905846e-05, | |
| "loss": 0.2814, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12399876117706299, | |
| "step": 1505, | |
| "valid_targets_mean": 3954.5, | |
| "valid_targets_min": 2001 | |
| }, | |
| { | |
| "epoch": 1.7481181239143022, | |
| "grad_norm": 0.4831572116294239, | |
| "learning_rate": 3.733875651265566e-05, | |
| "loss": 0.2871, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11142855882644653, | |
| "step": 1510, | |
| "valid_targets_mean": 5059.6, | |
| "valid_targets_min": 1767 | |
| }, | |
| { | |
| "epoch": 1.7539085118702953, | |
| "grad_norm": 0.5298635253583321, | |
| "learning_rate": 3.7309916740903904e-05, | |
| "loss": 0.2856, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1220664381980896, | |
| "step": 1515, | |
| "valid_targets_mean": 4714.0, | |
| "valid_targets_min": 1511 | |
| }, | |
| { | |
| "epoch": 1.7596988998262884, | |
| "grad_norm": 0.6941605178493575, | |
| "learning_rate": 3.728093280484039e-05, | |
| "loss": 0.2811, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13680237531661987, | |
| "step": 1520, | |
| "valid_targets_mean": 2861.0, | |
| "valid_targets_min": 974 | |
| }, | |
| { | |
| "epoch": 1.7654892877822814, | |
| "grad_norm": 0.6081050330394304, | |
| "learning_rate": 3.725180494585561e-05, | |
| "loss": 0.2891, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12764519453048706, | |
| "step": 1525, | |
| "valid_targets_mean": 3792.6, | |
| "valid_targets_min": 2339 | |
| }, | |
| { | |
| "epoch": 1.7712796757382745, | |
| "grad_norm": 0.5905461387221194, | |
| "learning_rate": 3.722253340653869e-05, | |
| "loss": 0.2836, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12954270839691162, | |
| "step": 1530, | |
| "valid_targets_mean": 3333.1, | |
| "valid_targets_min": 1249 | |
| }, | |
| { | |
| "epoch": 1.7770700636942676, | |
| "grad_norm": 0.6502080970822529, | |
| "learning_rate": 3.719311843067541e-05, | |
| "loss": 0.2941, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14891153573989868, | |
| "step": 1535, | |
| "valid_targets_mean": 3493.1, | |
| "valid_targets_min": 1223 | |
| }, | |
| { | |
| "epoch": 1.7828604516502606, | |
| "grad_norm": 0.6974244607751314, | |
| "learning_rate": 3.7163560263246126e-05, | |
| "loss": 0.3008, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15846528112888336, | |
| "step": 1540, | |
| "valid_targets_mean": 3285.8, | |
| "valid_targets_min": 1318 | |
| }, | |
| { | |
| "epoch": 1.7886508396062535, | |
| "grad_norm": 0.7366256261342894, | |
| "learning_rate": 3.713385915042376e-05, | |
| "loss": 0.2932, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13647408783435822, | |
| "step": 1545, | |
| "valid_targets_mean": 2070.5, | |
| "valid_targets_min": 1002 | |
| }, | |
| { | |
| "epoch": 1.7944412275622468, | |
| "grad_norm": 0.6302534919300808, | |
| "learning_rate": 3.710401533957177e-05, | |
| "loss": 0.3154, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13620512187480927, | |
| "step": 1550, | |
| "valid_targets_mean": 3328.1, | |
| "valid_targets_min": 845 | |
| }, | |
| { | |
| "epoch": 1.8002316155182396, | |
| "grad_norm": 0.5569959028822243, | |
| "learning_rate": 3.707402907924202e-05, | |
| "loss": 0.2843, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15685269236564636, | |
| "step": 1555, | |
| "valid_targets_mean": 4657.4, | |
| "valid_targets_min": 2747 | |
| }, | |
| { | |
| "epoch": 1.806022003474233, | |
| "grad_norm": 0.6254433967452091, | |
| "learning_rate": 3.704390061917279e-05, | |
| "loss": 0.3039, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1573985517024994, | |
| "step": 1560, | |
| "valid_targets_mean": 4166.4, | |
| "valid_targets_min": 1241 | |
| }, | |
| { | |
| "epoch": 1.8118123914302258, | |
| "grad_norm": 0.6605344075486596, | |
| "learning_rate": 3.701363021028663e-05, | |
| "loss": 0.3032, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16879332065582275, | |
| "step": 1565, | |
| "valid_targets_mean": 3567.1, | |
| "valid_targets_min": 992 | |
| }, | |
| { | |
| "epoch": 1.8176027793862188, | |
| "grad_norm": 0.5992245067450327, | |
| "learning_rate": 3.698321810468834e-05, | |
| "loss": 0.295, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1365610510110855, | |
| "step": 1570, | |
| "valid_targets_mean": 3343.5, | |
| "valid_targets_min": 1211 | |
| }, | |
| { | |
| "epoch": 1.823393167342212, | |
| "grad_norm": 0.6939069026436273, | |
| "learning_rate": 3.695266455566277e-05, | |
| "loss": 0.2982, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16110128164291382, | |
| "step": 1575, | |
| "valid_targets_mean": 3081.6, | |
| "valid_targets_min": 1889 | |
| }, | |
| { | |
| "epoch": 1.829183555298205, | |
| "grad_norm": 0.658296632984435, | |
| "learning_rate": 3.692196981767284e-05, | |
| "loss": 0.2953, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12838894128799438, | |
| "step": 1580, | |
| "valid_targets_mean": 3233.9, | |
| "valid_targets_min": 1624 | |
| }, | |
| { | |
| "epoch": 1.834973943254198, | |
| "grad_norm": 0.5932833946506312, | |
| "learning_rate": 3.689113414635729e-05, | |
| "loss": 0.297, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15206484496593475, | |
| "step": 1585, | |
| "valid_targets_mean": 3817.4, | |
| "valid_targets_min": 2763 | |
| }, | |
| { | |
| "epoch": 1.8407643312101911, | |
| "grad_norm": 0.6417579915172227, | |
| "learning_rate": 3.6860157798528655e-05, | |
| "loss": 0.2943, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1469709575176239, | |
| "step": 1590, | |
| "valid_targets_mean": 3037.0, | |
| "valid_targets_min": 1928 | |
| }, | |
| { | |
| "epoch": 1.8465547191661842, | |
| "grad_norm": 0.5635895780043579, | |
| "learning_rate": 3.6829041032171065e-05, | |
| "loss": 0.2761, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1479199230670929, | |
| "step": 1595, | |
| "valid_targets_mean": 4733.2, | |
| "valid_targets_min": 3147 | |
| }, | |
| { | |
| "epoch": 1.852345107122177, | |
| "grad_norm": 0.5927830627612026, | |
| "learning_rate": 3.679778410643812e-05, | |
| "loss": 0.2827, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12866196036338806, | |
| "step": 1600, | |
| "valid_targets_mean": 3348.4, | |
| "valid_targets_min": 1844 | |
| }, | |
| { | |
| "epoch": 1.8581354950781703, | |
| "grad_norm": 0.6165140209070116, | |
| "learning_rate": 3.676638728165072e-05, | |
| "loss": 0.2897, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14650055766105652, | |
| "step": 1605, | |
| "valid_targets_mean": 4360.2, | |
| "valid_targets_min": 2330 | |
| }, | |
| { | |
| "epoch": 1.8639258830341632, | |
| "grad_norm": 0.6817824524905477, | |
| "learning_rate": 3.673485081929491e-05, | |
| "loss": 0.2924, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1639326512813568, | |
| "step": 1610, | |
| "valid_targets_mean": 3574.1, | |
| "valid_targets_min": 900 | |
| }, | |
| { | |
| "epoch": 1.8697162709901565, | |
| "grad_norm": 0.5791877359098891, | |
| "learning_rate": 3.6703174982019696e-05, | |
| "loss": 0.2825, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15437737107276917, | |
| "step": 1615, | |
| "valid_targets_mean": 5092.8, | |
| "valid_targets_min": 1713 | |
| }, | |
| { | |
| "epoch": 1.8755066589461493, | |
| "grad_norm": 0.602627175280257, | |
| "learning_rate": 3.667136003363486e-05, | |
| "loss": 0.2986, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16277575492858887, | |
| "step": 1620, | |
| "valid_targets_mean": 4057.1, | |
| "valid_targets_min": 981 | |
| }, | |
| { | |
| "epoch": 1.8812970469021424, | |
| "grad_norm": 0.6112362273386295, | |
| "learning_rate": 3.663940623910874e-05, | |
| "loss": 0.2862, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1648990511894226, | |
| "step": 1625, | |
| "valid_targets_mean": 4127.9, | |
| "valid_targets_min": 2098 | |
| }, | |
| { | |
| "epoch": 1.8870874348581355, | |
| "grad_norm": 0.8037151230076643, | |
| "learning_rate": 3.660731386456607e-05, | |
| "loss": 0.2949, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1346992403268814, | |
| "step": 1630, | |
| "valid_targets_mean": 3101.1, | |
| "valid_targets_min": 1107 | |
| }, | |
| { | |
| "epoch": 1.8928778228141285, | |
| "grad_norm": 0.5804972757492853, | |
| "learning_rate": 3.6575083177285726e-05, | |
| "loss": 0.2863, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13580957055091858, | |
| "step": 1635, | |
| "valid_targets_mean": 3718.0, | |
| "valid_targets_min": 2273 | |
| }, | |
| { | |
| "epoch": 1.8986682107701216, | |
| "grad_norm": 0.5905918270634198, | |
| "learning_rate": 3.654271444569851e-05, | |
| "loss": 0.2809, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12466682493686676, | |
| "step": 1640, | |
| "valid_targets_mean": 3316.0, | |
| "valid_targets_min": 1920 | |
| }, | |
| { | |
| "epoch": 1.9044585987261147, | |
| "grad_norm": 0.5420420764028602, | |
| "learning_rate": 3.651020793938491e-05, | |
| "loss": 0.2699, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14009517431259155, | |
| "step": 1645, | |
| "valid_targets_mean": 4710.8, | |
| "valid_targets_min": 1687 | |
| }, | |
| { | |
| "epoch": 1.9102489866821077, | |
| "grad_norm": 0.6025644717103477, | |
| "learning_rate": 3.6477563929072876e-05, | |
| "loss": 0.2788, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12080429494380951, | |
| "step": 1650, | |
| "valid_targets_mean": 3075.8, | |
| "valid_targets_min": 1375 | |
| }, | |
| { | |
| "epoch": 1.9160393746381006, | |
| "grad_norm": 0.6012321362774695, | |
| "learning_rate": 3.6444782686635534e-05, | |
| "loss": 0.285, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12920139729976654, | |
| "step": 1655, | |
| "valid_targets_mean": 3776.4, | |
| "valid_targets_min": 906 | |
| }, | |
| { | |
| "epoch": 1.9218297625940939, | |
| "grad_norm": 0.6380750764288287, | |
| "learning_rate": 3.641186448508895e-05, | |
| "loss": 0.2832, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1428179144859314, | |
| "step": 1660, | |
| "valid_targets_mean": 3393.9, | |
| "valid_targets_min": 1874 | |
| }, | |
| { | |
| "epoch": 1.9276201505500867, | |
| "grad_norm": 0.5674294662426445, | |
| "learning_rate": 3.637880959858983e-05, | |
| "loss": 0.2753, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14175750315189362, | |
| "step": 1665, | |
| "valid_targets_mean": 3503.4, | |
| "valid_targets_min": 1877 | |
| }, | |
| { | |
| "epoch": 1.93341053850608, | |
| "grad_norm": 0.7553571883007889, | |
| "learning_rate": 3.634561830243326e-05, | |
| "loss": 0.2803, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13421311974525452, | |
| "step": 1670, | |
| "valid_targets_mean": 3744.4, | |
| "valid_targets_min": 1697 | |
| }, | |
| { | |
| "epoch": 1.9392009264620729, | |
| "grad_norm": 0.6443562711759654, | |
| "learning_rate": 3.631229087305039e-05, | |
| "loss": 0.28, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14671412110328674, | |
| "step": 1675, | |
| "valid_targets_mean": 3015.1, | |
| "valid_targets_min": 1679 | |
| }, | |
| { | |
| "epoch": 1.9449913144180662, | |
| "grad_norm": 0.6391854883175944, | |
| "learning_rate": 3.6278827588006175e-05, | |
| "loss": 0.3057, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14905647933483124, | |
| "step": 1680, | |
| "valid_targets_mean": 3292.0, | |
| "valid_targets_min": 1695 | |
| }, | |
| { | |
| "epoch": 1.950781702374059, | |
| "grad_norm": 0.6184510237966775, | |
| "learning_rate": 3.6245228725996994e-05, | |
| "loss": 0.2887, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12810581922531128, | |
| "step": 1685, | |
| "valid_targets_mean": 3148.8, | |
| "valid_targets_min": 2114 | |
| }, | |
| { | |
| "epoch": 1.956572090330052, | |
| "grad_norm": 0.6268276802626227, | |
| "learning_rate": 3.62114945668484e-05, | |
| "loss": 0.2842, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11713097244501114, | |
| "step": 1690, | |
| "valid_targets_mean": 2852.0, | |
| "valid_targets_min": 993 | |
| }, | |
| { | |
| "epoch": 1.9623624782860452, | |
| "grad_norm": 0.6314414968187866, | |
| "learning_rate": 3.617762539151275e-05, | |
| "loss": 0.2761, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1318410336971283, | |
| "step": 1695, | |
| "valid_targets_mean": 3059.8, | |
| "valid_targets_min": 1275 | |
| }, | |
| { | |
| "epoch": 1.9681528662420382, | |
| "grad_norm": 0.5568778782912174, | |
| "learning_rate": 3.6143621482066855e-05, | |
| "loss": 0.2821, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15411721169948578, | |
| "step": 1700, | |
| "valid_targets_mean": 4479.0, | |
| "valid_targets_min": 2250 | |
| }, | |
| { | |
| "epoch": 1.9739432541980313, | |
| "grad_norm": 0.5822106858978399, | |
| "learning_rate": 3.610948312170966e-05, | |
| "loss": 0.2833, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17367419600486755, | |
| "step": 1705, | |
| "valid_targets_mean": 4952.2, | |
| "valid_targets_min": 2892 | |
| }, | |
| { | |
| "epoch": 1.9797336421540244, | |
| "grad_norm": 0.5437001699478157, | |
| "learning_rate": 3.60752105947599e-05, | |
| "loss": 0.2951, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13590948283672333, | |
| "step": 1710, | |
| "valid_targets_mean": 4026.4, | |
| "valid_targets_min": 2935 | |
| }, | |
| { | |
| "epoch": 1.9855240301100174, | |
| "grad_norm": 0.5316143241569335, | |
| "learning_rate": 3.6040804186653664e-05, | |
| "loss": 0.2837, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13042062520980835, | |
| "step": 1715, | |
| "valid_targets_mean": 4085.2, | |
| "valid_targets_min": 1817 | |
| }, | |
| { | |
| "epoch": 1.9913144180660103, | |
| "grad_norm": 0.5673752637324689, | |
| "learning_rate": 3.600626418394208e-05, | |
| "loss": 0.2858, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14978323876857758, | |
| "step": 1720, | |
| "valid_targets_mean": 4050.4, | |
| "valid_targets_min": 1727 | |
| }, | |
| { | |
| "epoch": 1.9971048060220036, | |
| "grad_norm": 0.554230279363605, | |
| "learning_rate": 3.597159087428891e-05, | |
| "loss": 0.2833, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11575158685445786, | |
| "step": 1725, | |
| "valid_targets_mean": 3682.9, | |
| "valid_targets_min": 1622 | |
| }, | |
| { | |
| "epoch": 2.002316155182397, | |
| "grad_norm": 0.4286585060877845, | |
| "learning_rate": 3.593678454646815e-05, | |
| "loss": 0.2567, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10107292234897614, | |
| "step": 1730, | |
| "valid_targets_mean": 8514.0, | |
| "valid_targets_min": 6788 | |
| }, | |
| { | |
| "epoch": 2.0081065431383904, | |
| "grad_norm": 0.4700487542281608, | |
| "learning_rate": 3.5901845490361636e-05, | |
| "loss": 0.2299, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11241059750318527, | |
| "step": 1735, | |
| "valid_targets_mean": 6668.0, | |
| "valid_targets_min": 5933 | |
| }, | |
| { | |
| "epoch": 2.0138969310943833, | |
| "grad_norm": 0.5656127248804904, | |
| "learning_rate": 3.586677399695659e-05, | |
| "loss": 0.2208, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10180117189884186, | |
| "step": 1740, | |
| "valid_targets_mean": 4107.0, | |
| "valid_targets_min": 161 | |
| }, | |
| { | |
| "epoch": 2.0196873190503766, | |
| "grad_norm": 0.4107340656584201, | |
| "learning_rate": 3.583157035834327e-05, | |
| "loss": 0.2171, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09953369945287704, | |
| "step": 1745, | |
| "valid_targets_mean": 7946.0, | |
| "valid_targets_min": 5596 | |
| }, | |
| { | |
| "epoch": 2.0254777070063694, | |
| "grad_norm": 0.4169057084257508, | |
| "learning_rate": 3.579623486771247e-05, | |
| "loss": 0.2085, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1120513528585434, | |
| "step": 1750, | |
| "valid_targets_mean": 7392.8, | |
| "valid_targets_min": 5629 | |
| }, | |
| { | |
| "epoch": 2.0312680949623623, | |
| "grad_norm": 0.42108322176936563, | |
| "learning_rate": 3.576076781935311e-05, | |
| "loss": 0.2123, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10831199586391449, | |
| "step": 1755, | |
| "valid_targets_mean": 6879.6, | |
| "valid_targets_min": 5113 | |
| }, | |
| { | |
| "epoch": 2.0370584829183556, | |
| "grad_norm": 0.46891626874187786, | |
| "learning_rate": 3.572516950864977e-05, | |
| "loss": 0.2294, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11068766564130783, | |
| "step": 1760, | |
| "valid_targets_mean": 6919.5, | |
| "valid_targets_min": 5080 | |
| }, | |
| { | |
| "epoch": 2.0428488708743484, | |
| "grad_norm": 0.45738686416436514, | |
| "learning_rate": 3.568944023208026e-05, | |
| "loss": 0.2223, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09865154325962067, | |
| "step": 1765, | |
| "valid_targets_mean": 6377.0, | |
| "valid_targets_min": 3939 | |
| }, | |
| { | |
| "epoch": 2.0486392588303417, | |
| "grad_norm": 0.3953598703172133, | |
| "learning_rate": 3.565358028721311e-05, | |
| "loss": 0.2108, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10655589401721954, | |
| "step": 1770, | |
| "valid_targets_mean": 6392.2, | |
| "valid_targets_min": 4355 | |
| }, | |
| { | |
| "epoch": 2.0544296467863346, | |
| "grad_norm": 0.5105530013505218, | |
| "learning_rate": 3.5617589972705104e-05, | |
| "loss": 0.2056, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12021344900131226, | |
| "step": 1775, | |
| "valid_targets_mean": 5196.5, | |
| "valid_targets_min": 137 | |
| }, | |
| { | |
| "epoch": 2.060220034742328, | |
| "grad_norm": 0.3802473517239841, | |
| "learning_rate": 3.5581469588298826e-05, | |
| "loss": 0.2089, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09380616992712021, | |
| "step": 1780, | |
| "valid_targets_mean": 7344.4, | |
| "valid_targets_min": 5244 | |
| }, | |
| { | |
| "epoch": 2.0660104226983207, | |
| "grad_norm": 0.37526809344745743, | |
| "learning_rate": 3.554521943482011e-05, | |
| "loss": 0.2025, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08865103870630264, | |
| "step": 1785, | |
| "valid_targets_mean": 7921.0, | |
| "valid_targets_min": 5311 | |
| }, | |
| { | |
| "epoch": 2.071800810654314, | |
| "grad_norm": 0.42312765517314455, | |
| "learning_rate": 3.550883981417559e-05, | |
| "loss": 0.2125, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10647544264793396, | |
| "step": 1790, | |
| "valid_targets_mean": 7931.2, | |
| "valid_targets_min": 6015 | |
| }, | |
| { | |
| "epoch": 2.077591198610307, | |
| "grad_norm": 0.37102703581187674, | |
| "learning_rate": 3.547233102935013e-05, | |
| "loss": 0.2016, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09514321386814117, | |
| "step": 1795, | |
| "valid_targets_mean": 7640.8, | |
| "valid_targets_min": 5248 | |
| }, | |
| { | |
| "epoch": 2.0833815865663, | |
| "grad_norm": 0.3804558512027849, | |
| "learning_rate": 3.5435693384404346e-05, | |
| "loss": 0.2004, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0999484658241272, | |
| "step": 1800, | |
| "valid_targets_mean": 7249.6, | |
| "valid_targets_min": 5217 | |
| }, | |
| { | |
| "epoch": 2.089171974522293, | |
| "grad_norm": 0.39961483892181837, | |
| "learning_rate": 3.5398927184472044e-05, | |
| "loss": 0.2095, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.115473173558712, | |
| "step": 1805, | |
| "valid_targets_mean": 7619.9, | |
| "valid_targets_min": 5094 | |
| }, | |
| { | |
| "epoch": 2.0949623624782863, | |
| "grad_norm": 0.40322220155035593, | |
| "learning_rate": 3.536203273575769e-05, | |
| "loss": 0.2025, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09707570821046829, | |
| "step": 1810, | |
| "valid_targets_mean": 7732.8, | |
| "valid_targets_min": 5842 | |
| }, | |
| { | |
| "epoch": 2.100752750434279, | |
| "grad_norm": 0.4013736383136356, | |
| "learning_rate": 3.532501034553389e-05, | |
| "loss": 0.2051, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10486122965812683, | |
| "step": 1815, | |
| "valid_targets_mean": 7952.9, | |
| "valid_targets_min": 5494 | |
| }, | |
| { | |
| "epoch": 2.106543138390272, | |
| "grad_norm": 0.3976795627706171, | |
| "learning_rate": 3.528786032213875e-05, | |
| "loss": 0.2207, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10951173305511475, | |
| "step": 1820, | |
| "valid_targets_mean": 8001.0, | |
| "valid_targets_min": 5454 | |
| }, | |
| { | |
| "epoch": 2.1123335263462653, | |
| "grad_norm": 0.5536985357272681, | |
| "learning_rate": 3.525058297497339e-05, | |
| "loss": 0.227, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08318760991096497, | |
| "step": 1825, | |
| "valid_targets_mean": 2985.1, | |
| "valid_targets_min": 1896 | |
| }, | |
| { | |
| "epoch": 2.118123914302258, | |
| "grad_norm": 0.41262587121815264, | |
| "learning_rate": 3.521317861449935e-05, | |
| "loss": 0.1836, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08865408599376678, | |
| "step": 1830, | |
| "valid_targets_mean": 6714.6, | |
| "valid_targets_min": 4917 | |
| }, | |
| { | |
| "epoch": 2.1239143022582514, | |
| "grad_norm": 0.42576908508936934, | |
| "learning_rate": 3.517564755223596e-05, | |
| "loss": 0.2008, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11042719334363937, | |
| "step": 1835, | |
| "valid_targets_mean": 7595.2, | |
| "valid_targets_min": 5488 | |
| }, | |
| { | |
| "epoch": 2.1297046902142442, | |
| "grad_norm": 0.489870656589442, | |
| "learning_rate": 3.513799010075778e-05, | |
| "loss": 0.203, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10994037985801697, | |
| "step": 1840, | |
| "valid_targets_mean": 7046.1, | |
| "valid_targets_min": 5579 | |
| }, | |
| { | |
| "epoch": 2.1354950781702375, | |
| "grad_norm": 0.4438672691242358, | |
| "learning_rate": 3.5100206573692016e-05, | |
| "loss": 0.2105, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10180249065160751, | |
| "step": 1845, | |
| "valid_targets_mean": 7108.0, | |
| "valid_targets_min": 4963 | |
| }, | |
| { | |
| "epoch": 2.1412854661262304, | |
| "grad_norm": 0.42176761912224064, | |
| "learning_rate": 3.506229728571587e-05, | |
| "loss": 0.2057, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10390205681324005, | |
| "step": 1850, | |
| "valid_targets_mean": 6810.1, | |
| "valid_targets_min": 5666 | |
| }, | |
| { | |
| "epoch": 2.1470758540822237, | |
| "grad_norm": 0.4205231074425438, | |
| "learning_rate": 3.502426255255393e-05, | |
| "loss": 0.1977, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09647151827812195, | |
| "step": 1855, | |
| "valid_targets_mean": 8068.5, | |
| "valid_targets_min": 5524 | |
| }, | |
| { | |
| "epoch": 2.1528662420382165, | |
| "grad_norm": 0.39913914589973887, | |
| "learning_rate": 3.498610269097554e-05, | |
| "loss": 0.1801, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08500435948371887, | |
| "step": 1860, | |
| "valid_targets_mean": 6553.8, | |
| "valid_targets_min": 3621 | |
| }, | |
| { | |
| "epoch": 2.15865662999421, | |
| "grad_norm": 0.41904414215616076, | |
| "learning_rate": 3.49478180187922e-05, | |
| "loss": 0.1829, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09690581262111664, | |
| "step": 1865, | |
| "valid_targets_mean": 6762.5, | |
| "valid_targets_min": 4968 | |
| }, | |
| { | |
| "epoch": 2.1644470179502027, | |
| "grad_norm": 0.4556773026118942, | |
| "learning_rate": 3.490940885485484e-05, | |
| "loss": 0.1961, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09846589714288712, | |
| "step": 1870, | |
| "valid_targets_mean": 6143.9, | |
| "valid_targets_min": 4917 | |
| }, | |
| { | |
| "epoch": 2.1702374059061955, | |
| "grad_norm": 0.6116632442341243, | |
| "learning_rate": 3.487087551905123e-05, | |
| "loss": 0.2021, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13917821645736694, | |
| "step": 1875, | |
| "valid_targets_mean": 4583.8, | |
| "valid_targets_min": 215 | |
| }, | |
| { | |
| "epoch": 2.176027793862189, | |
| "grad_norm": 0.39727120128324056, | |
| "learning_rate": 3.483221833230331e-05, | |
| "loss": 0.2094, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10544075816869736, | |
| "step": 1880, | |
| "valid_targets_mean": 7317.1, | |
| "valid_targets_min": 4948 | |
| }, | |
| { | |
| "epoch": 2.1818181818181817, | |
| "grad_norm": 0.4817720608717964, | |
| "learning_rate": 3.4793437616564484e-05, | |
| "loss": 0.2238, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10634658485651016, | |
| "step": 1885, | |
| "valid_targets_mean": 6437.6, | |
| "valid_targets_min": 4510 | |
| }, | |
| { | |
| "epoch": 2.187608569774175, | |
| "grad_norm": 0.4479444586079385, | |
| "learning_rate": 3.4754533694816964e-05, | |
| "loss": 0.2146, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11554727703332901, | |
| "step": 1890, | |
| "valid_targets_mean": 7707.9, | |
| "valid_targets_min": 5796 | |
| }, | |
| { | |
| "epoch": 2.193398957730168, | |
| "grad_norm": 0.39456266062547346, | |
| "learning_rate": 3.471550689106907e-05, | |
| "loss": 0.2038, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09939637780189514, | |
| "step": 1895, | |
| "valid_targets_mean": 7900.9, | |
| "valid_targets_min": 5643 | |
| }, | |
| { | |
| "epoch": 2.199189345686161, | |
| "grad_norm": 0.4314337219750506, | |
| "learning_rate": 3.4676357530352544e-05, | |
| "loss": 0.2051, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11286012828350067, | |
| "step": 1900, | |
| "valid_targets_mean": 7627.2, | |
| "valid_targets_min": 5742 | |
| }, | |
| { | |
| "epoch": 2.204979733642154, | |
| "grad_norm": 0.39814070122791084, | |
| "learning_rate": 3.463708593871983e-05, | |
| "loss": 0.2088, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1185033842921257, | |
| "step": 1905, | |
| "valid_targets_mean": 8725.6, | |
| "valid_targets_min": 5141 | |
| }, | |
| { | |
| "epoch": 2.2107701215981472, | |
| "grad_norm": 0.4342438080398501, | |
| "learning_rate": 3.459769244324136e-05, | |
| "loss": 0.2061, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11473096907138824, | |
| "step": 1910, | |
| "valid_targets_mean": 6967.8, | |
| "valid_targets_min": 6088 | |
| }, | |
| { | |
| "epoch": 2.21656050955414, | |
| "grad_norm": 0.4064856622407791, | |
| "learning_rate": 3.455817737200283e-05, | |
| "loss": 0.2095, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0970906987786293, | |
| "step": 1915, | |
| "valid_targets_mean": 7140.1, | |
| "valid_targets_min": 4834 | |
| }, | |
| { | |
| "epoch": 2.2223508975101334, | |
| "grad_norm": 0.43104421279326466, | |
| "learning_rate": 3.451854105410248e-05, | |
| "loss": 0.2076, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.096015565097332, | |
| "step": 1920, | |
| "valid_targets_mean": 6706.9, | |
| "valid_targets_min": 5108 | |
| }, | |
| { | |
| "epoch": 2.228141285466126, | |
| "grad_norm": 0.47819652028982407, | |
| "learning_rate": 3.447878381964834e-05, | |
| "loss": 0.2107, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10358776897192001, | |
| "step": 1925, | |
| "valid_targets_mean": 5931.4, | |
| "valid_targets_min": 4708 | |
| }, | |
| { | |
| "epoch": 2.233931673422119, | |
| "grad_norm": 0.39514715435346154, | |
| "learning_rate": 3.4438905999755475e-05, | |
| "loss": 0.2022, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09589406102895737, | |
| "step": 1930, | |
| "valid_targets_mean": 7145.6, | |
| "valid_targets_min": 5280 | |
| }, | |
| { | |
| "epoch": 2.2397220613781124, | |
| "grad_norm": 0.7581947077221936, | |
| "learning_rate": 3.439890792654325e-05, | |
| "loss": 0.1974, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.06750276684761047, | |
| "step": 1935, | |
| "valid_targets_mean": 1554.9, | |
| "valid_targets_min": 592 | |
| }, | |
| { | |
| "epoch": 2.245512449334105, | |
| "grad_norm": 0.49252508143484086, | |
| "learning_rate": 3.435878993313255e-05, | |
| "loss": 0.1987, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11715425550937653, | |
| "step": 1940, | |
| "valid_targets_mean": 7069.5, | |
| "valid_targets_min": 4765 | |
| }, | |
| { | |
| "epoch": 2.2513028372900985, | |
| "grad_norm": 0.4648454181818097, | |
| "learning_rate": 3.4318552353642994e-05, | |
| "loss": 0.2036, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10164330899715424, | |
| "step": 1945, | |
| "valid_targets_mean": 6193.4, | |
| "valid_targets_min": 5587 | |
| }, | |
| { | |
| "epoch": 2.2570932252460914, | |
| "grad_norm": 0.43862119955162104, | |
| "learning_rate": 3.427819552319017e-05, | |
| "loss": 0.2059, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10625051707029343, | |
| "step": 1950, | |
| "valid_targets_mean": 6612.2, | |
| "valid_targets_min": 4894 | |
| }, | |
| { | |
| "epoch": 2.2628836132020846, | |
| "grad_norm": 0.42877814599863134, | |
| "learning_rate": 3.423771977788285e-05, | |
| "loss": 0.1977, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09473372995853424, | |
| "step": 1955, | |
| "valid_targets_mean": 6605.9, | |
| "valid_targets_min": 3795 | |
| }, | |
| { | |
| "epoch": 2.2686740011580775, | |
| "grad_norm": 0.42031608208073107, | |
| "learning_rate": 3.4197125454820155e-05, | |
| "loss": 0.2027, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10285250097513199, | |
| "step": 1960, | |
| "valid_targets_mean": 7186.8, | |
| "valid_targets_min": 4890 | |
| }, | |
| { | |
| "epoch": 2.274464389114071, | |
| "grad_norm": 0.4804930665105912, | |
| "learning_rate": 3.4156412892088795e-05, | |
| "loss": 0.2072, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10341675579547882, | |
| "step": 1965, | |
| "valid_targets_mean": 6544.4, | |
| "valid_targets_min": 4615 | |
| }, | |
| { | |
| "epoch": 2.2802547770700636, | |
| "grad_norm": 0.43505238303162125, | |
| "learning_rate": 3.4115582428760224e-05, | |
| "loss": 0.2083, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10975189507007599, | |
| "step": 1970, | |
| "valid_targets_mean": 6339.2, | |
| "valid_targets_min": 4798 | |
| }, | |
| { | |
| "epoch": 2.286045165026057, | |
| "grad_norm": 0.551542106228472, | |
| "learning_rate": 3.4074634404887814e-05, | |
| "loss": 0.1742, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1177409440279007, | |
| "step": 1975, | |
| "valid_targets_mean": 4558.8, | |
| "valid_targets_min": 148 | |
| }, | |
| { | |
| "epoch": 2.2918355529820498, | |
| "grad_norm": 0.4965808093480616, | |
| "learning_rate": 3.403356916150404e-05, | |
| "loss": 0.2041, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11211363971233368, | |
| "step": 1980, | |
| "valid_targets_mean": 6807.9, | |
| "valid_targets_min": 5164 | |
| }, | |
| { | |
| "epoch": 2.297625940938043, | |
| "grad_norm": 0.4554774656612426, | |
| "learning_rate": 3.3992387040617616e-05, | |
| "loss": 0.1972, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1016792431473732, | |
| "step": 1985, | |
| "valid_targets_mean": 7011.8, | |
| "valid_targets_min": 4974 | |
| }, | |
| { | |
| "epoch": 2.303416328894036, | |
| "grad_norm": 0.4493838431213323, | |
| "learning_rate": 3.395108838521068e-05, | |
| "loss": 0.2052, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09250965714454651, | |
| "step": 1990, | |
| "valid_targets_mean": 5794.2, | |
| "valid_targets_min": 4809 | |
| }, | |
| { | |
| "epoch": 2.3092067168500288, | |
| "grad_norm": 0.4323765965599609, | |
| "learning_rate": 3.390967353923591e-05, | |
| "loss": 0.1925, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09102576971054077, | |
| "step": 1995, | |
| "valid_targets_mean": 6401.8, | |
| "valid_targets_min": 4842 | |
| }, | |
| { | |
| "epoch": 2.314997104806022, | |
| "grad_norm": 0.41963893132513386, | |
| "learning_rate": 3.3868142847613677e-05, | |
| "loss": 0.1988, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10178063809871674, | |
| "step": 2000, | |
| "valid_targets_mean": 6182.1, | |
| "valid_targets_min": 5116 | |
| }, | |
| { | |
| "epoch": 2.320787492762015, | |
| "grad_norm": 0.4072833090612429, | |
| "learning_rate": 3.382649665622914e-05, | |
| "loss": 0.1985, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09133348613977432, | |
| "step": 2005, | |
| "valid_targets_mean": 6165.1, | |
| "valid_targets_min": 5764 | |
| }, | |
| { | |
| "epoch": 2.326577880718008, | |
| "grad_norm": 0.5814652127282149, | |
| "learning_rate": 3.378473531192942e-05, | |
| "loss": 0.2801, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16335105895996094, | |
| "step": 2010, | |
| "valid_targets_mean": 4151.0, | |
| "valid_targets_min": 1945 | |
| }, | |
| { | |
| "epoch": 2.332368268674001, | |
| "grad_norm": 0.5117966431044899, | |
| "learning_rate": 3.374285916252065e-05, | |
| "loss": 0.2763, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14518912136554718, | |
| "step": 2015, | |
| "valid_targets_mean": 6084.9, | |
| "valid_targets_min": 1799 | |
| }, | |
| { | |
| "epoch": 2.3381586566299943, | |
| "grad_norm": 0.5520391493076426, | |
| "learning_rate": 3.370086855676514e-05, | |
| "loss": 0.274, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16817054152488708, | |
| "step": 2020, | |
| "valid_targets_mean": 6738.0, | |
| "valid_targets_min": 3015 | |
| }, | |
| { | |
| "epoch": 2.343949044585987, | |
| "grad_norm": 0.5090140114428693, | |
| "learning_rate": 3.36587638443784e-05, | |
| "loss": 0.2702, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1441977322101593, | |
| "step": 2025, | |
| "valid_targets_mean": 5991.5, | |
| "valid_targets_min": 1681 | |
| }, | |
| { | |
| "epoch": 2.3497394325419805, | |
| "grad_norm": 0.5499747140361684, | |
| "learning_rate": 3.361654537602632e-05, | |
| "loss": 0.2763, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1109984964132309, | |
| "step": 2030, | |
| "valid_targets_mean": 3827.5, | |
| "valid_targets_min": 1044 | |
| }, | |
| { | |
| "epoch": 2.3555298204979733, | |
| "grad_norm": 0.5622165946749246, | |
| "learning_rate": 3.3574213503322146e-05, | |
| "loss": 0.2831, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1245369240641594, | |
| "step": 2035, | |
| "valid_targets_mean": 3889.4, | |
| "valid_targets_min": 1951 | |
| }, | |
| { | |
| "epoch": 2.361320208453966, | |
| "grad_norm": 0.5041215476638641, | |
| "learning_rate": 3.3531768578823646e-05, | |
| "loss": 0.2689, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12516576051712036, | |
| "step": 2040, | |
| "valid_targets_mean": 5299.0, | |
| "valid_targets_min": 2060 | |
| }, | |
| { | |
| "epoch": 2.3671105964099595, | |
| "grad_norm": 0.5007207114227822, | |
| "learning_rate": 3.348921095603011e-05, | |
| "loss": 0.269, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12523436546325684, | |
| "step": 2045, | |
| "valid_targets_mean": 5303.0, | |
| "valid_targets_min": 2424 | |
| }, | |
| { | |
| "epoch": 2.3729009843659528, | |
| "grad_norm": 0.5749076294108161, | |
| "learning_rate": 3.344654098937942e-05, | |
| "loss": 0.2844, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16740959882736206, | |
| "step": 2050, | |
| "valid_targets_mean": 5090.9, | |
| "valid_targets_min": 2077 | |
| }, | |
| { | |
| "epoch": 2.3786913723219456, | |
| "grad_norm": 0.6036836968287327, | |
| "learning_rate": 3.3403759034245117e-05, | |
| "loss": 0.2766, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12114214897155762, | |
| "step": 2055, | |
| "valid_targets_mean": 4620.9, | |
| "valid_targets_min": 2022 | |
| }, | |
| { | |
| "epoch": 2.3844817602779385, | |
| "grad_norm": 0.5375277920208905, | |
| "learning_rate": 3.336086544693344e-05, | |
| "loss": 0.2827, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15950371325016022, | |
| "step": 2060, | |
| "valid_targets_mean": 5276.0, | |
| "valid_targets_min": 1499 | |
| }, | |
| { | |
| "epoch": 2.3902721482339317, | |
| "grad_norm": 0.49513745930204944, | |
| "learning_rate": 3.331786058468032e-05, | |
| "loss": 0.2669, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1467110812664032, | |
| "step": 2065, | |
| "valid_targets_mean": 6256.1, | |
| "valid_targets_min": 1013 | |
| }, | |
| { | |
| "epoch": 2.3960625361899246, | |
| "grad_norm": 0.5783002445562354, | |
| "learning_rate": 3.327474480564846e-05, | |
| "loss": 0.2768, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12622880935668945, | |
| "step": 2070, | |
| "valid_targets_mean": 3896.4, | |
| "valid_targets_min": 1939 | |
| }, | |
| { | |
| "epoch": 2.401852924145918, | |
| "grad_norm": 0.5769722093369325, | |
| "learning_rate": 3.32315184689243e-05, | |
| "loss": 0.2663, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15862694382667542, | |
| "step": 2075, | |
| "valid_targets_mean": 4025.5, | |
| "valid_targets_min": 1820 | |
| }, | |
| { | |
| "epoch": 2.4076433121019107, | |
| "grad_norm": 0.6525497480217162, | |
| "learning_rate": 3.318818193451509e-05, | |
| "loss": 0.2843, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13170477747917175, | |
| "step": 2080, | |
| "valid_targets_mean": 2984.9, | |
| "valid_targets_min": 1589 | |
| }, | |
| { | |
| "epoch": 2.413433700057904, | |
| "grad_norm": 0.5784325309997747, | |
| "learning_rate": 3.31447355633458e-05, | |
| "loss": 0.2723, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12977421283721924, | |
| "step": 2085, | |
| "valid_targets_mean": 4404.4, | |
| "valid_targets_min": 2668 | |
| }, | |
| { | |
| "epoch": 2.419224088013897, | |
| "grad_norm": 0.6165233848368223, | |
| "learning_rate": 3.3101179717256214e-05, | |
| "loss": 0.2765, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.156073197722435, | |
| "step": 2090, | |
| "valid_targets_mean": 3997.1, | |
| "valid_targets_min": 2530 | |
| }, | |
| { | |
| "epoch": 2.42501447596989, | |
| "grad_norm": 0.66620163280296, | |
| "learning_rate": 3.305751475899783e-05, | |
| "loss": 0.2863, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.132912278175354, | |
| "step": 2095, | |
| "valid_targets_mean": 3860.4, | |
| "valid_targets_min": 2170 | |
| }, | |
| { | |
| "epoch": 2.430804863925883, | |
| "grad_norm": 0.6585893355253127, | |
| "learning_rate": 3.3013741052230915e-05, | |
| "loss": 0.283, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09839099645614624, | |
| "step": 2100, | |
| "valid_targets_mean": 2333.4, | |
| "valid_targets_min": 1128 | |
| }, | |
| { | |
| "epoch": 2.436595251881876, | |
| "grad_norm": 0.6194017217583334, | |
| "learning_rate": 3.2969858961521426e-05, | |
| "loss": 0.277, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14759422838687897, | |
| "step": 2105, | |
| "valid_targets_mean": 3412.4, | |
| "valid_targets_min": 2001 | |
| }, | |
| { | |
| "epoch": 2.442385639837869, | |
| "grad_norm": 0.6290180566419377, | |
| "learning_rate": 3.292586885233797e-05, | |
| "loss": 0.2794, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1449579894542694, | |
| "step": 2110, | |
| "valid_targets_mean": 3916.6, | |
| "valid_targets_min": 1267 | |
| }, | |
| { | |
| "epoch": 2.448176027793862, | |
| "grad_norm": 0.6830786628543166, | |
| "learning_rate": 3.288177109104879e-05, | |
| "loss": 0.2682, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1411552131175995, | |
| "step": 2115, | |
| "valid_targets_mean": 3868.5, | |
| "valid_targets_min": 2234 | |
| }, | |
| { | |
| "epoch": 2.4539664157498553, | |
| "grad_norm": 0.6523431302307824, | |
| "learning_rate": 3.2837566044918726e-05, | |
| "loss": 0.2542, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1432904601097107, | |
| "step": 2120, | |
| "valid_targets_mean": 3154.4, | |
| "valid_targets_min": 1403 | |
| }, | |
| { | |
| "epoch": 2.459756803705848, | |
| "grad_norm": 0.6315305472107878, | |
| "learning_rate": 3.2793254082106086e-05, | |
| "loss": 0.2735, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15610270202159882, | |
| "step": 2125, | |
| "valid_targets_mean": 4107.5, | |
| "valid_targets_min": 1588 | |
| }, | |
| { | |
| "epoch": 2.4655471916618414, | |
| "grad_norm": 0.6560897876305025, | |
| "learning_rate": 3.274883557165965e-05, | |
| "loss": 0.2797, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15769946575164795, | |
| "step": 2130, | |
| "valid_targets_mean": 3702.9, | |
| "valid_targets_min": 1902 | |
| }, | |
| { | |
| "epoch": 2.4713375796178343, | |
| "grad_norm": 0.5244545462951082, | |
| "learning_rate": 3.270431088351558e-05, | |
| "loss": 0.2673, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12711435556411743, | |
| "step": 2135, | |
| "valid_targets_mean": 4432.1, | |
| "valid_targets_min": 2171 | |
| }, | |
| { | |
| "epoch": 2.4771279675738276, | |
| "grad_norm": 0.6242616772853598, | |
| "learning_rate": 3.265968038849433e-05, | |
| "loss": 0.27, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16617780923843384, | |
| "step": 2140, | |
| "valid_targets_mean": 4749.9, | |
| "valid_targets_min": 2677 | |
| }, | |
| { | |
| "epoch": 2.4829183555298204, | |
| "grad_norm": 0.6069853123821018, | |
| "learning_rate": 3.261494445829752e-05, | |
| "loss": 0.2614, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1648349165916443, | |
| "step": 2145, | |
| "valid_targets_mean": 4314.9, | |
| "valid_targets_min": 2725 | |
| }, | |
| { | |
| "epoch": 2.4887087434858137, | |
| "grad_norm": 0.7004803993263157, | |
| "learning_rate": 3.257010346550493e-05, | |
| "loss": 0.2782, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17026188969612122, | |
| "step": 2150, | |
| "valid_targets_mean": 3132.1, | |
| "valid_targets_min": 1685 | |
| }, | |
| { | |
| "epoch": 2.4944991314418066, | |
| "grad_norm": 0.5664274832522125, | |
| "learning_rate": 3.252515778357131e-05, | |
| "loss": 0.2565, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1272943615913391, | |
| "step": 2155, | |
| "valid_targets_mean": 4106.8, | |
| "valid_targets_min": 2384 | |
| }, | |
| { | |
| "epoch": 2.5002895193978, | |
| "grad_norm": 0.5996482260516004, | |
| "learning_rate": 3.248010778682333e-05, | |
| "loss": 0.266, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14419519901275635, | |
| "step": 2160, | |
| "valid_targets_mean": 4175.2, | |
| "valid_targets_min": 2060 | |
| }, | |
| { | |
| "epoch": 2.5060799073537927, | |
| "grad_norm": 0.6708401852808833, | |
| "learning_rate": 3.2434953850456426e-05, | |
| "loss": 0.2651, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13591818511486053, | |
| "step": 2165, | |
| "valid_targets_mean": 3389.1, | |
| "valid_targets_min": 1458 | |
| }, | |
| { | |
| "epoch": 2.5118702953097856, | |
| "grad_norm": 0.6366477783226726, | |
| "learning_rate": 3.238969635053168e-05, | |
| "loss": 0.2705, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1480705738067627, | |
| "step": 2170, | |
| "valid_targets_mean": 4349.9, | |
| "valid_targets_min": 1512 | |
| }, | |
| { | |
| "epoch": 2.517660683265779, | |
| "grad_norm": 0.6905097825586773, | |
| "learning_rate": 3.234433566397271e-05, | |
| "loss": 0.2621, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11380015313625336, | |
| "step": 2175, | |
| "valid_targets_mean": 2479.8, | |
| "valid_targets_min": 1480 | |
| }, | |
| { | |
| "epoch": 2.5234510712217717, | |
| "grad_norm": 0.6197924359922682, | |
| "learning_rate": 3.229887216856251e-05, | |
| "loss": 0.2715, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13098078966140747, | |
| "step": 2180, | |
| "valid_targets_mean": 3904.8, | |
| "valid_targets_min": 1338 | |
| }, | |
| { | |
| "epoch": 2.529241459177765, | |
| "grad_norm": 0.5762184284428006, | |
| "learning_rate": 3.2253306242940306e-05, | |
| "loss": 0.2706, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1326790302991867, | |
| "step": 2185, | |
| "valid_targets_mean": 3684.2, | |
| "valid_targets_min": 1276 | |
| }, | |
| { | |
| "epoch": 2.535031847133758, | |
| "grad_norm": 0.635981487933494, | |
| "learning_rate": 3.22076382665984e-05, | |
| "loss": 0.2668, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12587392330169678, | |
| "step": 2190, | |
| "valid_targets_mean": 4008.1, | |
| "valid_targets_min": 1991 | |
| }, | |
| { | |
| "epoch": 2.540822235089751, | |
| "grad_norm": 0.6365019756761532, | |
| "learning_rate": 3.216186861987905e-05, | |
| "loss": 0.2634, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14298924803733826, | |
| "step": 2195, | |
| "valid_targets_mean": 3331.6, | |
| "valid_targets_min": 924 | |
| }, | |
| { | |
| "epoch": 2.546612623045744, | |
| "grad_norm": 0.6169273093000665, | |
| "learning_rate": 3.211599768397121e-05, | |
| "loss": 0.2693, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13531437516212463, | |
| "step": 2200, | |
| "valid_targets_mean": 4542.5, | |
| "valid_targets_min": 2089 | |
| }, | |
| { | |
| "epoch": 2.5524030110017373, | |
| "grad_norm": 0.5905221275607936, | |
| "learning_rate": 3.2070025840907474e-05, | |
| "loss": 0.2661, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11663927137851715, | |
| "step": 2205, | |
| "valid_targets_mean": 3307.2, | |
| "valid_targets_min": 1643 | |
| }, | |
| { | |
| "epoch": 2.55819339895773, | |
| "grad_norm": 0.5867800821675817, | |
| "learning_rate": 3.202395347356079e-05, | |
| "loss": 0.2622, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13914506137371063, | |
| "step": 2210, | |
| "valid_targets_mean": 3690.4, | |
| "valid_targets_min": 1244 | |
| }, | |
| { | |
| "epoch": 2.563983786913723, | |
| "grad_norm": 0.568570395278458, | |
| "learning_rate": 3.1977780965641334e-05, | |
| "loss": 0.2719, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14504513144493103, | |
| "step": 2215, | |
| "valid_targets_mean": 4846.2, | |
| "valid_targets_min": 2710 | |
| }, | |
| { | |
| "epoch": 2.5697741748697163, | |
| "grad_norm": 0.6412931743418574, | |
| "learning_rate": 3.193150870169329e-05, | |
| "loss": 0.2665, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1039309948682785, | |
| "step": 2220, | |
| "valid_targets_mean": 2848.8, | |
| "valid_targets_min": 977 | |
| }, | |
| { | |
| "epoch": 2.5755645628257096, | |
| "grad_norm": 0.5723133537467101, | |
| "learning_rate": 3.188513706709165e-05, | |
| "loss": 0.2641, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14200431108474731, | |
| "step": 2225, | |
| "valid_targets_mean": 4040.8, | |
| "valid_targets_min": 810 | |
| }, | |
| { | |
| "epoch": 2.5813549507817024, | |
| "grad_norm": 0.5793947108625295, | |
| "learning_rate": 3.183866644803901e-05, | |
| "loss": 0.2479, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1353072226047516, | |
| "step": 2230, | |
| "valid_targets_mean": 3452.1, | |
| "valid_targets_min": 1798 | |
| }, | |
| { | |
| "epoch": 2.5871453387376953, | |
| "grad_norm": 0.9774813400849247, | |
| "learning_rate": 3.1792097231562364e-05, | |
| "loss": 0.2719, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1235002651810646, | |
| "step": 2235, | |
| "valid_targets_mean": 3939.5, | |
| "valid_targets_min": 2035 | |
| }, | |
| { | |
| "epoch": 2.5929357266936885, | |
| "grad_norm": 0.5872365776559091, | |
| "learning_rate": 3.174542980550984e-05, | |
| "loss": 0.2515, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12853936851024628, | |
| "step": 2240, | |
| "valid_targets_mean": 3559.2, | |
| "valid_targets_min": 1261 | |
| }, | |
| { | |
| "epoch": 2.5987261146496814, | |
| "grad_norm": 0.6816129912425755, | |
| "learning_rate": 3.1698664558547515e-05, | |
| "loss": 0.2725, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1662760227918625, | |
| "step": 2245, | |
| "valid_targets_mean": 3653.4, | |
| "valid_targets_min": 1714 | |
| }, | |
| { | |
| "epoch": 2.6045165026056747, | |
| "grad_norm": 0.6388954275905896, | |
| "learning_rate": 3.165180188015616e-05, | |
| "loss": 0.2599, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10945677757263184, | |
| "step": 2250, | |
| "valid_targets_mean": 3019.4, | |
| "valid_targets_min": 1219 | |
| }, | |
| { | |
| "epoch": 2.6103068905616675, | |
| "grad_norm": 0.6210381392513558, | |
| "learning_rate": 3.160484216062798e-05, | |
| "loss": 0.2698, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11807023733854294, | |
| "step": 2255, | |
| "valid_targets_mean": 3603.2, | |
| "valid_targets_min": 2058 | |
| }, | |
| { | |
| "epoch": 2.616097278517661, | |
| "grad_norm": 0.6448257402388333, | |
| "learning_rate": 3.1557785791063406e-05, | |
| "loss": 0.2616, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10894700884819031, | |
| "step": 2260, | |
| "valid_targets_mean": 2962.5, | |
| "valid_targets_min": 916 | |
| }, | |
| { | |
| "epoch": 2.6218876664736537, | |
| "grad_norm": 0.5926187115075681, | |
| "learning_rate": 3.1510633163367786e-05, | |
| "loss": 0.2747, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11035982519388199, | |
| "step": 2265, | |
| "valid_targets_mean": 3654.0, | |
| "valid_targets_min": 1302 | |
| }, | |
| { | |
| "epoch": 2.627678054429647, | |
| "grad_norm": 0.6634112051804333, | |
| "learning_rate": 3.146338467024816e-05, | |
| "loss": 0.2496, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13553813099861145, | |
| "step": 2270, | |
| "valid_targets_mean": 3304.9, | |
| "valid_targets_min": 1278 | |
| }, | |
| { | |
| "epoch": 2.63346844238564, | |
| "grad_norm": 0.5932977731351348, | |
| "learning_rate": 3.141604070520998e-05, | |
| "loss": 0.2515, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1372648924589157, | |
| "step": 2275, | |
| "valid_targets_mean": 3934.6, | |
| "valid_targets_min": 2471 | |
| }, | |
| { | |
| "epoch": 2.6392588303416327, | |
| "grad_norm": 0.5629032334171726, | |
| "learning_rate": 3.136860166255382e-05, | |
| "loss": 0.2435, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13328105211257935, | |
| "step": 2280, | |
| "valid_targets_mean": 4429.2, | |
| "valid_targets_min": 1922 | |
| }, | |
| { | |
| "epoch": 2.645049218297626, | |
| "grad_norm": 0.5869812911680073, | |
| "learning_rate": 3.132106793737209e-05, | |
| "loss": 0.2707, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14404839277267456, | |
| "step": 2285, | |
| "valid_targets_mean": 3957.9, | |
| "valid_targets_min": 2672 | |
| }, | |
| { | |
| "epoch": 2.6508396062536193, | |
| "grad_norm": 0.6229051741089654, | |
| "learning_rate": 3.127343992554577e-05, | |
| "loss": 0.2578, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1221669465303421, | |
| "step": 2290, | |
| "valid_targets_mean": 3290.0, | |
| "valid_targets_min": 1086 | |
| }, | |
| { | |
| "epoch": 2.656629994209612, | |
| "grad_norm": 0.5835055556088607, | |
| "learning_rate": 3.122571802374112e-05, | |
| "loss": 0.2573, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1405826210975647, | |
| "step": 2295, | |
| "valid_targets_mean": 5206.5, | |
| "valid_targets_min": 1566 | |
| }, | |
| { | |
| "epoch": 2.662420382165605, | |
| "grad_norm": 0.5863525815985173, | |
| "learning_rate": 3.117790262940631e-05, | |
| "loss": 0.2602, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14586801826953888, | |
| "step": 2300, | |
| "valid_targets_mean": 3793.8, | |
| "valid_targets_min": 2010 | |
| }, | |
| { | |
| "epoch": 2.6682107701215982, | |
| "grad_norm": 0.6796155332616456, | |
| "learning_rate": 3.112999414076818e-05, | |
| "loss": 0.2711, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1526874303817749, | |
| "step": 2305, | |
| "valid_targets_mean": 3481.2, | |
| "valid_targets_min": 1548 | |
| }, | |
| { | |
| "epoch": 2.674001158077591, | |
| "grad_norm": 0.6658592899639356, | |
| "learning_rate": 3.1081992956828905e-05, | |
| "loss": 0.2612, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1481824517250061, | |
| "step": 2310, | |
| "valid_targets_mean": 3473.6, | |
| "valid_targets_min": 1352 | |
| }, | |
| { | |
| "epoch": 2.6797915460335844, | |
| "grad_norm": 0.5785234103831691, | |
| "learning_rate": 3.103389947736266e-05, | |
| "loss": 0.2533, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12380445748567581, | |
| "step": 2315, | |
| "valid_targets_mean": 4093.6, | |
| "valid_targets_min": 1483 | |
| }, | |
| { | |
| "epoch": 2.6855819339895772, | |
| "grad_norm": 0.5915985888499322, | |
| "learning_rate": 3.098571410291228e-05, | |
| "loss": 0.2476, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11295592784881592, | |
| "step": 2320, | |
| "valid_targets_mean": 3756.0, | |
| "valid_targets_min": 2015 | |
| }, | |
| { | |
| "epoch": 2.69137232194557, | |
| "grad_norm": 0.5559748529158701, | |
| "learning_rate": 3.093743723478598e-05, | |
| "loss": 0.2521, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12268182635307312, | |
| "step": 2325, | |
| "valid_targets_mean": 4435.0, | |
| "valid_targets_min": 1678 | |
| }, | |
| { | |
| "epoch": 2.6971627099015634, | |
| "grad_norm": 0.6503040030028072, | |
| "learning_rate": 3.0889069275053916e-05, | |
| "loss": 0.2537, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12161822617053986, | |
| "step": 2330, | |
| "valid_targets_mean": 2909.4, | |
| "valid_targets_min": 1568 | |
| }, | |
| { | |
| "epoch": 2.7029530978575567, | |
| "grad_norm": 0.5637056236595298, | |
| "learning_rate": 3.084061062654495e-05, | |
| "loss": 0.2436, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11895494163036346, | |
| "step": 2335, | |
| "valid_targets_mean": 3716.4, | |
| "valid_targets_min": 1282 | |
| }, | |
| { | |
| "epoch": 2.7087434858135495, | |
| "grad_norm": 0.6519618594888245, | |
| "learning_rate": 3.0792061692843196e-05, | |
| "loss": 0.243, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.131838858127594, | |
| "step": 2340, | |
| "valid_targets_mean": 3126.6, | |
| "valid_targets_min": 1312 | |
| }, | |
| { | |
| "epoch": 2.7145338737695424, | |
| "grad_norm": 0.6028390887407604, | |
| "learning_rate": 3.074342287828473e-05, | |
| "loss": 0.2429, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12123403698205948, | |
| "step": 2345, | |
| "valid_targets_mean": 3599.6, | |
| "valid_targets_min": 1764 | |
| }, | |
| { | |
| "epoch": 2.7203242617255357, | |
| "grad_norm": 0.6838931794867641, | |
| "learning_rate": 3.069469458795418e-05, | |
| "loss": 0.2692, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11124719679355621, | |
| "step": 2350, | |
| "valid_targets_mean": 2447.4, | |
| "valid_targets_min": 1175 | |
| }, | |
| { | |
| "epoch": 2.7261146496815285, | |
| "grad_norm": 0.5398446575139603, | |
| "learning_rate": 3.064587722768136e-05, | |
| "loss": 0.2613, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13309673964977264, | |
| "step": 2355, | |
| "valid_targets_mean": 4826.4, | |
| "valid_targets_min": 2546 | |
| }, | |
| { | |
| "epoch": 2.731905037637522, | |
| "grad_norm": 0.7287355653313029, | |
| "learning_rate": 3.059697120403791e-05, | |
| "loss": 0.2552, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1539398431777954, | |
| "step": 2360, | |
| "valid_targets_mean": 3297.0, | |
| "valid_targets_min": 1244 | |
| }, | |
| { | |
| "epoch": 2.7376954255935146, | |
| "grad_norm": 0.6110697593914238, | |
| "learning_rate": 3.05479769243339e-05, | |
| "loss": 0.2566, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14367569983005524, | |
| "step": 2365, | |
| "valid_targets_mean": 4389.5, | |
| "valid_targets_min": 1501 | |
| }, | |
| { | |
| "epoch": 2.743485813549508, | |
| "grad_norm": 0.5904899994841355, | |
| "learning_rate": 3.049889479661441e-05, | |
| "loss": 0.2505, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12652909755706787, | |
| "step": 2370, | |
| "valid_targets_mean": 3887.8, | |
| "valid_targets_min": 2097 | |
| }, | |
| { | |
| "epoch": 2.749276201505501, | |
| "grad_norm": 0.5293444804964664, | |
| "learning_rate": 3.044972522965618e-05, | |
| "loss": 0.2488, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10903643071651459, | |
| "step": 2375, | |
| "valid_targets_mean": 4151.8, | |
| "valid_targets_min": 1558 | |
| }, | |
| { | |
| "epoch": 2.755066589461494, | |
| "grad_norm": 0.5345690566260798, | |
| "learning_rate": 3.040046863296416e-05, | |
| "loss": 0.2511, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11641837656497955, | |
| "step": 2380, | |
| "valid_targets_mean": 4776.1, | |
| "valid_targets_min": 1145 | |
| }, | |
| { | |
| "epoch": 2.760856977417487, | |
| "grad_norm": 0.6345446481949528, | |
| "learning_rate": 3.0351125416768155e-05, | |
| "loss": 0.2523, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1530282199382782, | |
| "step": 2385, | |
| "valid_targets_mean": 3919.2, | |
| "valid_targets_min": 1428 | |
| }, | |
| { | |
| "epoch": 2.7666473653734798, | |
| "grad_norm": 0.5969763049450747, | |
| "learning_rate": 3.0301695992019317e-05, | |
| "loss": 0.2525, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14372681081295013, | |
| "step": 2390, | |
| "valid_targets_mean": 4515.1, | |
| "valid_targets_min": 2510 | |
| }, | |
| { | |
| "epoch": 2.772437753329473, | |
| "grad_norm": 0.6250162114900181, | |
| "learning_rate": 3.0252180770386843e-05, | |
| "loss": 0.249, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1120104044675827, | |
| "step": 2395, | |
| "valid_targets_mean": 3257.1, | |
| "valid_targets_min": 1301 | |
| }, | |
| { | |
| "epoch": 2.7782281412854664, | |
| "grad_norm": 0.6445992710336376, | |
| "learning_rate": 3.020258016425443e-05, | |
| "loss": 0.2672, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13314387202262878, | |
| "step": 2400, | |
| "valid_targets_mean": 3663.1, | |
| "valid_targets_min": 1971 | |
| }, | |
| { | |
| "epoch": 2.784018529241459, | |
| "grad_norm": 0.6483103155518952, | |
| "learning_rate": 3.0152894586716935e-05, | |
| "loss": 0.264, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1609051674604416, | |
| "step": 2405, | |
| "valid_targets_mean": 3979.0, | |
| "valid_targets_min": 1054 | |
| }, | |
| { | |
| "epoch": 2.789808917197452, | |
| "grad_norm": 0.6156498937088388, | |
| "learning_rate": 3.0103124451576855e-05, | |
| "loss": 0.2565, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1333981156349182, | |
| "step": 2410, | |
| "valid_targets_mean": 4076.8, | |
| "valid_targets_min": 2206 | |
| }, | |
| { | |
| "epoch": 2.7955993051534453, | |
| "grad_norm": 0.5862048353117455, | |
| "learning_rate": 3.0053270173340946e-05, | |
| "loss": 0.2754, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12760062515735626, | |
| "step": 2415, | |
| "valid_targets_mean": 3744.9, | |
| "valid_targets_min": 721 | |
| }, | |
| { | |
| "epoch": 2.801389693109438, | |
| "grad_norm": 0.5830520407651704, | |
| "learning_rate": 3.000333216721674e-05, | |
| "loss": 0.2542, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1376606523990631, | |
| "step": 2420, | |
| "valid_targets_mean": 4235.1, | |
| "valid_targets_min": 1625 | |
| }, | |
| { | |
| "epoch": 2.8071800810654315, | |
| "grad_norm": 0.6455391194137454, | |
| "learning_rate": 2.9953310849109073e-05, | |
| "loss": 0.2701, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14893652498722076, | |
| "step": 2425, | |
| "valid_targets_mean": 4103.1, | |
| "valid_targets_min": 2589 | |
| }, | |
| { | |
| "epoch": 2.8129704690214243, | |
| "grad_norm": 0.7062520885378663, | |
| "learning_rate": 2.9903206635616656e-05, | |
| "loss": 0.2711, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12403398752212524, | |
| "step": 2430, | |
| "valid_targets_mean": 2978.0, | |
| "valid_targets_min": 998 | |
| }, | |
| { | |
| "epoch": 2.8187608569774176, | |
| "grad_norm": 0.6206595353746295, | |
| "learning_rate": 2.9853019944028583e-05, | |
| "loss": 0.2556, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12284980714321136, | |
| "step": 2435, | |
| "valid_targets_mean": 4591.8, | |
| "valid_targets_min": 1186 | |
| }, | |
| { | |
| "epoch": 2.8245512449334105, | |
| "grad_norm": 0.5622116346549204, | |
| "learning_rate": 2.9802751192320855e-05, | |
| "loss": 0.2596, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11684159189462662, | |
| "step": 2440, | |
| "valid_targets_mean": 3200.6, | |
| "valid_targets_min": 1195 | |
| }, | |
| { | |
| "epoch": 2.8303416328894038, | |
| "grad_norm": 0.5998859708099616, | |
| "learning_rate": 2.9752400799152916e-05, | |
| "loss": 0.262, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11318564414978027, | |
| "step": 2445, | |
| "valid_targets_mean": 3047.4, | |
| "valid_targets_min": 1863 | |
| }, | |
| { | |
| "epoch": 2.8361320208453966, | |
| "grad_norm": 0.6486787196368696, | |
| "learning_rate": 2.9701969183864137e-05, | |
| "loss": 0.2666, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1260605901479721, | |
| "step": 2450, | |
| "valid_targets_mean": 3483.8, | |
| "valid_targets_min": 1483 | |
| }, | |
| { | |
| "epoch": 2.8419224088013895, | |
| "grad_norm": 0.6601872798879775, | |
| "learning_rate": 2.965145676647036e-05, | |
| "loss": 0.2592, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1035212054848671, | |
| "step": 2455, | |
| "valid_targets_mean": 3146.8, | |
| "valid_targets_min": 1662 | |
| }, | |
| { | |
| "epoch": 2.8477127967573828, | |
| "grad_norm": 0.5953589784047442, | |
| "learning_rate": 2.9600863967660368e-05, | |
| "loss": 0.2435, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1442669928073883, | |
| "step": 2460, | |
| "valid_targets_mean": 4545.2, | |
| "valid_targets_min": 2659 | |
| }, | |
| { | |
| "epoch": 2.853503184713376, | |
| "grad_norm": 0.6766549577613615, | |
| "learning_rate": 2.95501912087924e-05, | |
| "loss": 0.2541, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13332010805606842, | |
| "step": 2465, | |
| "valid_targets_mean": 3529.8, | |
| "valid_targets_min": 1434 | |
| }, | |
| { | |
| "epoch": 2.859293572669369, | |
| "grad_norm": 0.6037530300473921, | |
| "learning_rate": 2.9499438911890644e-05, | |
| "loss": 0.2498, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11366799473762512, | |
| "step": 2470, | |
| "valid_targets_mean": 2998.2, | |
| "valid_targets_min": 1074 | |
| }, | |
| { | |
| "epoch": 2.8650839606253617, | |
| "grad_norm": 0.6536835630432294, | |
| "learning_rate": 2.944860749964169e-05, | |
| "loss": 0.2668, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12703803181648254, | |
| "step": 2475, | |
| "valid_targets_mean": 3511.4, | |
| "valid_targets_min": 1510 | |
| }, | |
| { | |
| "epoch": 2.870874348581355, | |
| "grad_norm": 0.5995340016895321, | |
| "learning_rate": 2.939769739539106e-05, | |
| "loss": 0.2536, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1322011649608612, | |
| "step": 2480, | |
| "valid_targets_mean": 4068.1, | |
| "valid_targets_min": 1885 | |
| }, | |
| { | |
| "epoch": 2.876664736537348, | |
| "grad_norm": 0.60589407553132, | |
| "learning_rate": 2.934670902313964e-05, | |
| "loss": 0.2619, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12155967950820923, | |
| "step": 2485, | |
| "valid_targets_mean": 4412.6, | |
| "valid_targets_min": 3139 | |
| }, | |
| { | |
| "epoch": 2.882455124493341, | |
| "grad_norm": 0.6433580210305632, | |
| "learning_rate": 2.929564280754018e-05, | |
| "loss": 0.2537, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15175604820251465, | |
| "step": 2490, | |
| "valid_targets_mean": 3801.2, | |
| "valid_targets_min": 2598 | |
| }, | |
| { | |
| "epoch": 2.888245512449334, | |
| "grad_norm": 0.6383076431224991, | |
| "learning_rate": 2.924449917389373e-05, | |
| "loss": 0.2585, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11478400230407715, | |
| "step": 2495, | |
| "valid_targets_mean": 2971.0, | |
| "valid_targets_min": 1460 | |
| }, | |
| { | |
| "epoch": 2.8940359004053273, | |
| "grad_norm": 0.6281639839698074, | |
| "learning_rate": 2.919327854814611e-05, | |
| "loss": 0.2576, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1256842315196991, | |
| "step": 2500, | |
| "valid_targets_mean": 3538.9, | |
| "valid_targets_min": 1084 | |
| }, | |
| { | |
| "epoch": 2.89982628836132, | |
| "grad_norm": 0.6110228223154738, | |
| "learning_rate": 2.9141981356884367e-05, | |
| "loss": 0.2474, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13257163763046265, | |
| "step": 2505, | |
| "valid_targets_mean": 3759.2, | |
| "valid_targets_min": 1618 | |
| }, | |
| { | |
| "epoch": 2.9056166763173135, | |
| "grad_norm": 0.65313781157454, | |
| "learning_rate": 2.909060802733322e-05, | |
| "loss": 0.2427, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13252943754196167, | |
| "step": 2510, | |
| "valid_targets_mean": 3248.8, | |
| "valid_targets_min": 1193 | |
| }, | |
| { | |
| "epoch": 2.9114070642733063, | |
| "grad_norm": 0.5869454134091099, | |
| "learning_rate": 2.90391589873515e-05, | |
| "loss": 0.2467, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12256360799074173, | |
| "step": 2515, | |
| "valid_targets_mean": 4340.0, | |
| "valid_targets_min": 1564 | |
| }, | |
| { | |
| "epoch": 2.917197452229299, | |
| "grad_norm": 0.6284591952877979, | |
| "learning_rate": 2.8987634665428578e-05, | |
| "loss": 0.2505, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14555668830871582, | |
| "step": 2520, | |
| "valid_targets_mean": 4032.2, | |
| "valid_targets_min": 1472 | |
| }, | |
| { | |
| "epoch": 2.9229878401852925, | |
| "grad_norm": 0.5602935140755988, | |
| "learning_rate": 2.8936035490680817e-05, | |
| "loss": 0.2465, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10094116628170013, | |
| "step": 2525, | |
| "valid_targets_mean": 4291.0, | |
| "valid_targets_min": 1246 | |
| }, | |
| { | |
| "epoch": 2.9287782281412853, | |
| "grad_norm": 0.6333361634147205, | |
| "learning_rate": 2.8884361892847974e-05, | |
| "loss": 0.2403, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10485686361789703, | |
| "step": 2530, | |
| "valid_targets_mean": 3467.9, | |
| "valid_targets_min": 1724 | |
| }, | |
| { | |
| "epoch": 2.9345686160972786, | |
| "grad_norm": 0.6347563240307721, | |
| "learning_rate": 2.8832614302289644e-05, | |
| "loss": 0.2539, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13671885430812836, | |
| "step": 2535, | |
| "valid_targets_mean": 3630.6, | |
| "valid_targets_min": 1502 | |
| }, | |
| { | |
| "epoch": 2.9403590040532714, | |
| "grad_norm": 0.5703161738148554, | |
| "learning_rate": 2.8780793149981652e-05, | |
| "loss": 0.2467, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1201121062040329, | |
| "step": 2540, | |
| "valid_targets_mean": 4141.1, | |
| "valid_targets_min": 2134 | |
| }, | |
| { | |
| "epoch": 2.9461493920092647, | |
| "grad_norm": 0.6173130579998207, | |
| "learning_rate": 2.8728898867512493e-05, | |
| "loss": 0.2755, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11488363146781921, | |
| "step": 2545, | |
| "valid_targets_mean": 3016.4, | |
| "valid_targets_min": 2271 | |
| }, | |
| { | |
| "epoch": 2.9519397799652576, | |
| "grad_norm": 0.5508192014943009, | |
| "learning_rate": 2.86769318870797e-05, | |
| "loss": 0.2486, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13369382917881012, | |
| "step": 2550, | |
| "valid_targets_mean": 4785.1, | |
| "valid_targets_min": 2713 | |
| }, | |
| { | |
| "epoch": 2.957730167921251, | |
| "grad_norm": 0.6406012334203832, | |
| "learning_rate": 2.8624892641486294e-05, | |
| "loss": 0.247, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12603960931301117, | |
| "step": 2555, | |
| "valid_targets_mean": 3951.6, | |
| "valid_targets_min": 2373 | |
| }, | |
| { | |
| "epoch": 2.9635205558772437, | |
| "grad_norm": 0.6116187219621803, | |
| "learning_rate": 2.8572781564137116e-05, | |
| "loss": 0.2469, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11695685982704163, | |
| "step": 2560, | |
| "valid_targets_mean": 3756.1, | |
| "valid_targets_min": 2440 | |
| }, | |
| { | |
| "epoch": 2.9693109438332366, | |
| "grad_norm": 0.6592558498956914, | |
| "learning_rate": 2.852059908903529e-05, | |
| "loss": 0.2579, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1306173950433731, | |
| "step": 2565, | |
| "valid_targets_mean": 3366.6, | |
| "valid_targets_min": 2113 | |
| }, | |
| { | |
| "epoch": 2.97510133178923, | |
| "grad_norm": 0.6351786945052718, | |
| "learning_rate": 2.8468345650778527e-05, | |
| "loss": 0.2501, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10195891559123993, | |
| "step": 2570, | |
| "valid_targets_mean": 2892.4, | |
| "valid_targets_min": 757 | |
| }, | |
| { | |
| "epoch": 2.980891719745223, | |
| "grad_norm": 0.6244909255403139, | |
| "learning_rate": 2.8416021684555593e-05, | |
| "loss": 0.2569, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1276107132434845, | |
| "step": 2575, | |
| "valid_targets_mean": 4595.0, | |
| "valid_targets_min": 1928 | |
| }, | |
| { | |
| "epoch": 2.986682107701216, | |
| "grad_norm": 0.6168677811966271, | |
| "learning_rate": 2.836362762614262e-05, | |
| "loss": 0.257, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1577070951461792, | |
| "step": 2580, | |
| "valid_targets_mean": 4214.5, | |
| "valid_targets_min": 1800 | |
| }, | |
| { | |
| "epoch": 2.992472495657209, | |
| "grad_norm": 0.6036284614314452, | |
| "learning_rate": 2.8311163911899486e-05, | |
| "loss": 0.2502, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12150377035140991, | |
| "step": 2585, | |
| "valid_targets_mean": 3766.9, | |
| "valid_targets_min": 2922 | |
| }, | |
| { | |
| "epoch": 2.998262883613202, | |
| "grad_norm": 0.6497477524082161, | |
| "learning_rate": 2.8258630978766208e-05, | |
| "loss": 0.2541, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14457978308200836, | |
| "step": 2590, | |
| "valid_targets_mean": 3857.8, | |
| "valid_targets_min": 1214 | |
| }, | |
| { | |
| "epoch": 3.0034742327735957, | |
| "grad_norm": 0.5490993236724281, | |
| "learning_rate": 2.820602926425929e-05, | |
| "loss": 0.2168, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10193323343992233, | |
| "step": 2595, | |
| "valid_targets_mean": 5435.1, | |
| "valid_targets_min": 3516 | |
| }, | |
| { | |
| "epoch": 3.009264620729589, | |
| "grad_norm": 0.42556536290513536, | |
| "learning_rate": 2.815335920646807e-05, | |
| "loss": 0.2055, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0997970774769783, | |
| "step": 2600, | |
| "valid_targets_mean": 7654.2, | |
| "valid_targets_min": 6354 | |
| }, | |
| { | |
| "epoch": 3.015055008685582, | |
| "grad_norm": 0.42912438060333435, | |
| "learning_rate": 2.810062124405107e-05, | |
| "loss": 0.1991, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1131233498454094, | |
| "step": 2605, | |
| "valid_targets_mean": 7341.6, | |
| "valid_targets_min": 5754 | |
| }, | |
| { | |
| "epoch": 3.020845396641575, | |
| "grad_norm": 0.4117296149802389, | |
| "learning_rate": 2.8047815816232363e-05, | |
| "loss": 0.1935, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08657995611429214, | |
| "step": 2610, | |
| "valid_targets_mean": 7666.2, | |
| "valid_targets_min": 5666 | |
| }, | |
| { | |
| "epoch": 3.026635784597568, | |
| "grad_norm": 0.42744506674088706, | |
| "learning_rate": 2.7994943362797906e-05, | |
| "loss": 0.1957, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10150963068008423, | |
| "step": 2615, | |
| "valid_targets_mean": 7469.0, | |
| "valid_targets_min": 5646 | |
| }, | |
| { | |
| "epoch": 3.0324261725535613, | |
| "grad_norm": 0.49056118722907033, | |
| "learning_rate": 2.794200432409185e-05, | |
| "loss": 0.1951, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11447674036026001, | |
| "step": 2620, | |
| "valid_targets_mean": 7046.9, | |
| "valid_targets_min": 5571 | |
| }, | |
| { | |
| "epoch": 3.038216560509554, | |
| "grad_norm": 0.44039068925534824, | |
| "learning_rate": 2.788899914101292e-05, | |
| "loss": 0.2123, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11457256972789764, | |
| "step": 2625, | |
| "valid_targets_mean": 7827.6, | |
| "valid_targets_min": 5405 | |
| }, | |
| { | |
| "epoch": 3.044006948465547, | |
| "grad_norm": 0.41349146941679255, | |
| "learning_rate": 2.7835928255010713e-05, | |
| "loss": 0.1978, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10325706750154495, | |
| "step": 2630, | |
| "valid_targets_mean": 7023.5, | |
| "valid_targets_min": 5664 | |
| }, | |
| { | |
| "epoch": 3.0497973364215403, | |
| "grad_norm": 0.4593882416922647, | |
| "learning_rate": 2.778279210808203e-05, | |
| "loss": 0.1915, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0934155136346817, | |
| "step": 2635, | |
| "valid_targets_mean": 7069.8, | |
| "valid_targets_min": 5103 | |
| }, | |
| { | |
| "epoch": 3.055587724377533, | |
| "grad_norm": 0.4196669159756002, | |
| "learning_rate": 2.7729591142767175e-05, | |
| "loss": 0.1808, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09234358370304108, | |
| "step": 2640, | |
| "valid_targets_mean": 6712.4, | |
| "valid_targets_min": 5469 | |
| }, | |
| { | |
| "epoch": 3.0613781123335264, | |
| "grad_norm": 0.38396037776881536, | |
| "learning_rate": 2.7676325802146306e-05, | |
| "loss": 0.193, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09183961898088455, | |
| "step": 2645, | |
| "valid_targets_mean": 7289.2, | |
| "valid_targets_min": 5513 | |
| }, | |
| { | |
| "epoch": 3.0671685002895193, | |
| "grad_norm": 0.4087622775220975, | |
| "learning_rate": 2.762299652983573e-05, | |
| "loss": 0.1848, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09124772250652313, | |
| "step": 2650, | |
| "valid_targets_mean": 7022.1, | |
| "valid_targets_min": 5691 | |
| }, | |
| { | |
| "epoch": 3.0729588882455126, | |
| "grad_norm": 0.4131668084826757, | |
| "learning_rate": 2.756960376998418e-05, | |
| "loss": 0.1946, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09983032941818237, | |
| "step": 2655, | |
| "valid_targets_mean": 6736.0, | |
| "valid_targets_min": 5550 | |
| }, | |
| { | |
| "epoch": 3.0787492762015054, | |
| "grad_norm": 0.408189467149742, | |
| "learning_rate": 2.7516147967269163e-05, | |
| "loss": 0.1811, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0885000228881836, | |
| "step": 2660, | |
| "valid_targets_mean": 6346.9, | |
| "valid_targets_min": 5067 | |
| }, | |
| { | |
| "epoch": 3.0845396641574987, | |
| "grad_norm": 0.4030983371858413, | |
| "learning_rate": 2.746262956689322e-05, | |
| "loss": 0.1852, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0960773229598999, | |
| "step": 2665, | |
| "valid_targets_mean": 6784.0, | |
| "valid_targets_min": 5618 | |
| }, | |
| { | |
| "epoch": 3.0903300521134915, | |
| "grad_norm": 0.40401695487165545, | |
| "learning_rate": 2.7409049014580244e-05, | |
| "loss": 0.1932, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1078563779592514, | |
| "step": 2670, | |
| "valid_targets_mean": 7568.4, | |
| "valid_targets_min": 5856 | |
| }, | |
| { | |
| "epoch": 3.096120440069485, | |
| "grad_norm": 0.39748706417143886, | |
| "learning_rate": 2.7355406756571752e-05, | |
| "loss": 0.1845, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0892835333943367, | |
| "step": 2675, | |
| "valid_targets_mean": 7198.5, | |
| "valid_targets_min": 4793 | |
| }, | |
| { | |
| "epoch": 3.1019108280254777, | |
| "grad_norm": 0.4046911052941595, | |
| "learning_rate": 2.7301703239623152e-05, | |
| "loss": 0.1903, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09586253762245178, | |
| "step": 2680, | |
| "valid_targets_mean": 6680.5, | |
| "valid_targets_min": 5350 | |
| }, | |
| { | |
| "epoch": 3.1077012159814705, | |
| "grad_norm": 0.42963147337235336, | |
| "learning_rate": 2.7247938911000074e-05, | |
| "loss": 0.2027, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09971575438976288, | |
| "step": 2685, | |
| "valid_targets_mean": 6770.6, | |
| "valid_targets_min": 5753 | |
| }, | |
| { | |
| "epoch": 3.113491603937464, | |
| "grad_norm": 0.7914296289583164, | |
| "learning_rate": 2.7194114218474584e-05, | |
| "loss": 0.2, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07213065773248672, | |
| "step": 2690, | |
| "valid_targets_mean": 1419.8, | |
| "valid_targets_min": 289 | |
| }, | |
| { | |
| "epoch": 3.1192819918934567, | |
| "grad_norm": 0.4415378325560089, | |
| "learning_rate": 2.7140229610321517e-05, | |
| "loss": 0.1656, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0980992242693901, | |
| "step": 2695, | |
| "valid_targets_mean": 6784.2, | |
| "valid_targets_min": 5042 | |
| }, | |
| { | |
| "epoch": 3.12507237984945, | |
| "grad_norm": 0.39534955301481367, | |
| "learning_rate": 2.7086285535314686e-05, | |
| "loss": 0.1829, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09090964496135712, | |
| "step": 2700, | |
| "valid_targets_mean": 6774.9, | |
| "valid_targets_min": 3875 | |
| }, | |
| { | |
| "epoch": 3.130862767805443, | |
| "grad_norm": 0.45932881826395955, | |
| "learning_rate": 2.703228244272317e-05, | |
| "loss": 0.1882, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09747149050235748, | |
| "step": 2705, | |
| "valid_targets_mean": 7047.6, | |
| "valid_targets_min": 6134 | |
| }, | |
| { | |
| "epoch": 3.136653155761436, | |
| "grad_norm": 0.42200901741517993, | |
| "learning_rate": 2.6978220782307575e-05, | |
| "loss": 0.1922, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10043708980083466, | |
| "step": 2710, | |
| "valid_targets_mean": 6886.6, | |
| "valid_targets_min": 5869 | |
| }, | |
| { | |
| "epoch": 3.142443543717429, | |
| "grad_norm": 0.37704547310224273, | |
| "learning_rate": 2.6924101004316295e-05, | |
| "loss": 0.1864, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08716057986021042, | |
| "step": 2715, | |
| "valid_targets_mean": 7559.8, | |
| "valid_targets_min": 6145 | |
| }, | |
| { | |
| "epoch": 3.1482339316734222, | |
| "grad_norm": 0.3836760511290801, | |
| "learning_rate": 2.6869923559481743e-05, | |
| "loss": 0.1763, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08523589372634888, | |
| "step": 2720, | |
| "valid_targets_mean": 8920.5, | |
| "valid_targets_min": 4780 | |
| }, | |
| { | |
| "epoch": 3.154024319629415, | |
| "grad_norm": 0.348547170294601, | |
| "learning_rate": 2.68156888990166e-05, | |
| "loss": 0.1641, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07461922615766525, | |
| "step": 2725, | |
| "valid_targets_mean": 7393.4, | |
| "valid_targets_min": 5116 | |
| }, | |
| { | |
| "epoch": 3.1598147075854084, | |
| "grad_norm": 0.3728882684422235, | |
| "learning_rate": 2.6761397474610073e-05, | |
| "loss": 0.1673, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0829453319311142, | |
| "step": 2730, | |
| "valid_targets_mean": 8477.5, | |
| "valid_targets_min": 4033 | |
| }, | |
| { | |
| "epoch": 3.1656050955414012, | |
| "grad_norm": 0.41948282922444835, | |
| "learning_rate": 2.6707049738424117e-05, | |
| "loss": 0.187, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09367874264717102, | |
| "step": 2735, | |
| "valid_targets_mean": 6736.6, | |
| "valid_targets_min": 4646 | |
| }, | |
| { | |
| "epoch": 3.1713954834973945, | |
| "grad_norm": 0.43769731812052803, | |
| "learning_rate": 2.665264614308968e-05, | |
| "loss": 0.1778, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09644682705402374, | |
| "step": 2740, | |
| "valid_targets_mean": 7551.5, | |
| "valid_targets_min": 5461 | |
| }, | |
| { | |
| "epoch": 3.1771858714533874, | |
| "grad_norm": 0.4406770439909477, | |
| "learning_rate": 2.6598187141702923e-05, | |
| "loss": 0.1963, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11089968681335449, | |
| "step": 2745, | |
| "valid_targets_mean": 7091.0, | |
| "valid_targets_min": 4879 | |
| }, | |
| { | |
| "epoch": 3.1829762594093802, | |
| "grad_norm": 0.4515001973225758, | |
| "learning_rate": 2.6543673187821456e-05, | |
| "loss": 0.2016, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0904221385717392, | |
| "step": 2750, | |
| "valid_targets_mean": 6269.8, | |
| "valid_targets_min": 4606 | |
| }, | |
| { | |
| "epoch": 3.1887666473653735, | |
| "grad_norm": 0.4363784946771639, | |
| "learning_rate": 2.6489104735460562e-05, | |
| "loss": 0.1984, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08854199945926666, | |
| "step": 2755, | |
| "valid_targets_mean": 6818.4, | |
| "valid_targets_min": 5525 | |
| }, | |
| { | |
| "epoch": 3.1945570353213664, | |
| "grad_norm": 0.417191450993402, | |
| "learning_rate": 2.6434482239089398e-05, | |
| "loss": 0.1847, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09214088320732117, | |
| "step": 2760, | |
| "valid_targets_mean": 7365.4, | |
| "valid_targets_min": 5000 | |
| }, | |
| { | |
| "epoch": 3.2003474232773597, | |
| "grad_norm": 0.4337487632877825, | |
| "learning_rate": 2.637980615362723e-05, | |
| "loss": 0.191, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.084344781935215, | |
| "step": 2765, | |
| "valid_targets_mean": 6212.1, | |
| "valid_targets_min": 3988 | |
| }, | |
| { | |
| "epoch": 3.2061378112333525, | |
| "grad_norm": 0.3954975501331284, | |
| "learning_rate": 2.6325076934439633e-05, | |
| "loss": 0.1911, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09320168197154999, | |
| "step": 2770, | |
| "valid_targets_mean": 8396.9, | |
| "valid_targets_min": 6131 | |
| }, | |
| { | |
| "epoch": 3.211928199189346, | |
| "grad_norm": 0.4500355751831564, | |
| "learning_rate": 2.6270295037334713e-05, | |
| "loss": 0.1899, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09574490040540695, | |
| "step": 2775, | |
| "valid_targets_mean": 7651.0, | |
| "valid_targets_min": 4061 | |
| }, | |
| { | |
| "epoch": 3.2177185871453386, | |
| "grad_norm": 0.3938156686393294, | |
| "learning_rate": 2.6215460918559283e-05, | |
| "loss": 0.1932, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09371134638786316, | |
| "step": 2780, | |
| "valid_targets_mean": 7090.1, | |
| "valid_targets_min": 5203 | |
| }, | |
| { | |
| "epoch": 3.223508975101332, | |
| "grad_norm": 0.38793490038984524, | |
| "learning_rate": 2.6160575034795087e-05, | |
| "loss": 0.1919, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08991052210330963, | |
| "step": 2785, | |
| "valid_targets_mean": 7001.0, | |
| "valid_targets_min": 5362 | |
| }, | |
| { | |
| "epoch": 3.229299363057325, | |
| "grad_norm": 0.4173092770685844, | |
| "learning_rate": 2.6105637843155004e-05, | |
| "loss": 0.1926, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09407326579093933, | |
| "step": 2790, | |
| "valid_targets_mean": 6359.2, | |
| "valid_targets_min": 5236 | |
| }, | |
| { | |
| "epoch": 3.235089751013318, | |
| "grad_norm": 0.45548569764559765, | |
| "learning_rate": 2.60506498011792e-05, | |
| "loss": 0.1898, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10315366834402084, | |
| "step": 2795, | |
| "valid_targets_mean": 5939.8, | |
| "valid_targets_min": 4693 | |
| }, | |
| { | |
| "epoch": 3.240880138969311, | |
| "grad_norm": 1.238986424920992, | |
| "learning_rate": 2.599561136683136e-05, | |
| "loss": 0.1628, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.036311253905296326, | |
| "step": 2800, | |
| "valid_targets_mean": 208.6, | |
| "valid_targets_min": 134 | |
| }, | |
| { | |
| "epoch": 3.2466705269253042, | |
| "grad_norm": 0.5146729187215359, | |
| "learning_rate": 2.5940522998494863e-05, | |
| "loss": 0.1927, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09998352825641632, | |
| "step": 2805, | |
| "valid_targets_mean": 6008.5, | |
| "valid_targets_min": 4956 | |
| }, | |
| { | |
| "epoch": 3.252460914881297, | |
| "grad_norm": 0.42168821423053354, | |
| "learning_rate": 2.5885385154968954e-05, | |
| "loss": 0.1841, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08223581314086914, | |
| "step": 2810, | |
| "valid_targets_mean": 6042.0, | |
| "valid_targets_min": 4817 | |
| }, | |
| { | |
| "epoch": 3.25825130283729, | |
| "grad_norm": 0.4535711701413484, | |
| "learning_rate": 2.5830198295464923e-05, | |
| "loss": 0.1928, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09312181919813156, | |
| "step": 2815, | |
| "valid_targets_mean": 6608.6, | |
| "valid_targets_min": 5102 | |
| }, | |
| { | |
| "epoch": 3.264041690793283, | |
| "grad_norm": 0.4257103730536391, | |
| "learning_rate": 2.5774962879602285e-05, | |
| "loss": 0.1823, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09401199221611023, | |
| "step": 2820, | |
| "valid_targets_mean": 6670.1, | |
| "valid_targets_min": 4869 | |
| }, | |
| { | |
| "epoch": 3.269832078749276, | |
| "grad_norm": 0.44522510586129893, | |
| "learning_rate": 2.5719679367404955e-05, | |
| "loss": 0.1842, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08575493097305298, | |
| "step": 2825, | |
| "valid_targets_mean": 6088.4, | |
| "valid_targets_min": 4946 | |
| }, | |
| { | |
| "epoch": 3.2756224667052694, | |
| "grad_norm": 0.44406409728050933, | |
| "learning_rate": 2.5664348219297432e-05, | |
| "loss": 0.1927, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08885006606578827, | |
| "step": 2830, | |
| "valid_targets_mean": 6228.4, | |
| "valid_targets_min": 4856 | |
| }, | |
| { | |
| "epoch": 3.281412854661262, | |
| "grad_norm": 0.44972634761371566, | |
| "learning_rate": 2.560896989610093e-05, | |
| "loss": 0.1913, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09148620069026947, | |
| "step": 2835, | |
| "valid_targets_mean": 5376.6, | |
| "valid_targets_min": 4145 | |
| }, | |
| { | |
| "epoch": 3.2872032426172555, | |
| "grad_norm": 0.4931251776724191, | |
| "learning_rate": 2.555354485902954e-05, | |
| "loss": 0.1557, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09443989396095276, | |
| "step": 2840, | |
| "valid_targets_mean": 6309.2, | |
| "valid_targets_min": 5305 | |
| }, | |
| { | |
| "epoch": 3.2929936305732483, | |
| "grad_norm": 0.4301848434471346, | |
| "learning_rate": 2.5498073569686444e-05, | |
| "loss": 0.1845, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08982059359550476, | |
| "step": 2845, | |
| "valid_targets_mean": 6221.5, | |
| "valid_targets_min": 4636 | |
| }, | |
| { | |
| "epoch": 3.2987840185292416, | |
| "grad_norm": 0.41333674292252215, | |
| "learning_rate": 2.5442556490060014e-05, | |
| "loss": 0.1856, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09725256264209747, | |
| "step": 2850, | |
| "valid_targets_mean": 6831.5, | |
| "valid_targets_min": 5668 | |
| }, | |
| { | |
| "epoch": 3.3045744064852345, | |
| "grad_norm": 0.44385746477258825, | |
| "learning_rate": 2.5386994082519972e-05, | |
| "loss": 0.1871, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.097258061170578, | |
| "step": 2855, | |
| "valid_targets_mean": 6437.1, | |
| "valid_targets_min": 5004 | |
| }, | |
| { | |
| "epoch": 3.3103647944412273, | |
| "grad_norm": 0.41491139403789895, | |
| "learning_rate": 2.5331386809813565e-05, | |
| "loss": 0.1779, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08634775131940842, | |
| "step": 2860, | |
| "valid_targets_mean": 6205.8, | |
| "valid_targets_min": 5341 | |
| }, | |
| { | |
| "epoch": 3.3161551823972206, | |
| "grad_norm": 0.4154409527575692, | |
| "learning_rate": 2.5275735135061696e-05, | |
| "loss": 0.1839, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0978240966796875, | |
| "step": 2865, | |
| "valid_targets_mean": 6748.0, | |
| "valid_targets_min": 5363 | |
| }, | |
| { | |
| "epoch": 3.3219455703532135, | |
| "grad_norm": 0.453739504532685, | |
| "learning_rate": 2.5220039521755056e-05, | |
| "loss": 0.1827, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09334269911050797, | |
| "step": 2870, | |
| "valid_targets_mean": 5549.8, | |
| "valid_targets_min": 1320 | |
| }, | |
| { | |
| "epoch": 3.3277359583092068, | |
| "grad_norm": 0.5449848364061605, | |
| "learning_rate": 2.5164300433750287e-05, | |
| "loss": 0.2661, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1472756564617157, | |
| "step": 2875, | |
| "valid_targets_mean": 5766.9, | |
| "valid_targets_min": 746 | |
| }, | |
| { | |
| "epoch": 3.3335263462651996, | |
| "grad_norm": 0.5140311062323946, | |
| "learning_rate": 2.5108518335266104e-05, | |
| "loss": 0.2488, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13583004474639893, | |
| "step": 2880, | |
| "valid_targets_mean": 7107.5, | |
| "valid_targets_min": 3874 | |
| }, | |
| { | |
| "epoch": 3.339316734221193, | |
| "grad_norm": 0.49982562987867923, | |
| "learning_rate": 2.505269369087941e-05, | |
| "loss": 0.2419, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11187659204006195, | |
| "step": 2885, | |
| "valid_targets_mean": 5893.2, | |
| "valid_targets_min": 1321 | |
| }, | |
| { | |
| "epoch": 3.3451071221771858, | |
| "grad_norm": 0.5240008941141455, | |
| "learning_rate": 2.499682696552149e-05, | |
| "loss": 0.2427, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13496187329292297, | |
| "step": 2890, | |
| "valid_targets_mean": 6733.1, | |
| "valid_targets_min": 2373 | |
| }, | |
| { | |
| "epoch": 3.350897510133179, | |
| "grad_norm": 0.5844435666177288, | |
| "learning_rate": 2.4940918624474068e-05, | |
| "loss": 0.2558, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13430002331733704, | |
| "step": 2895, | |
| "valid_targets_mean": 4825.4, | |
| "valid_targets_min": 2745 | |
| }, | |
| { | |
| "epoch": 3.356687898089172, | |
| "grad_norm": 0.5368504900216172, | |
| "learning_rate": 2.488496913336546e-05, | |
| "loss": 0.2471, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11826002597808838, | |
| "step": 2900, | |
| "valid_targets_mean": 5337.6, | |
| "valid_targets_min": 1089 | |
| }, | |
| { | |
| "epoch": 3.362478286045165, | |
| "grad_norm": 0.5812257689605878, | |
| "learning_rate": 2.482897895816671e-05, | |
| "loss": 0.2444, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10885634273290634, | |
| "step": 2905, | |
| "valid_targets_mean": 3264.4, | |
| "valid_targets_min": 627 | |
| }, | |
| { | |
| "epoch": 3.368268674001158, | |
| "grad_norm": 0.5854102363595578, | |
| "learning_rate": 2.477294856518769e-05, | |
| "loss": 0.2512, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14446288347244263, | |
| "step": 2910, | |
| "valid_targets_mean": 4381.6, | |
| "valid_targets_min": 2605 | |
| }, | |
| { | |
| "epoch": 3.3740590619571513, | |
| "grad_norm": 0.5417087622693278, | |
| "learning_rate": 2.4716878421073224e-05, | |
| "loss": 0.2517, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13980954885482788, | |
| "step": 2915, | |
| "valid_targets_mean": 5933.9, | |
| "valid_targets_min": 1899 | |
| }, | |
| { | |
| "epoch": 3.379849449913144, | |
| "grad_norm": 0.5465882725499704, | |
| "learning_rate": 2.46607689927992e-05, | |
| "loss": 0.2482, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.128583624958992, | |
| "step": 2920, | |
| "valid_targets_mean": 4483.6, | |
| "valid_targets_min": 2747 | |
| }, | |
| { | |
| "epoch": 3.385639837869137, | |
| "grad_norm": 0.5592367972407115, | |
| "learning_rate": 2.460462074766868e-05, | |
| "loss": 0.2482, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12015688419342041, | |
| "step": 2925, | |
| "valid_targets_mean": 5165.4, | |
| "valid_targets_min": 2144 | |
| }, | |
| { | |
| "epoch": 3.3914302258251303, | |
| "grad_norm": 0.5816389165036916, | |
| "learning_rate": 2.4548434153308007e-05, | |
| "loss": 0.2432, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11116787046194077, | |
| "step": 2930, | |
| "valid_targets_mean": 4143.4, | |
| "valid_targets_min": 1461 | |
| }, | |
| { | |
| "epoch": 3.397220613781123, | |
| "grad_norm": 0.6260987341214326, | |
| "learning_rate": 2.4492209677662923e-05, | |
| "loss": 0.2468, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10702268779277802, | |
| "step": 2935, | |
| "valid_targets_mean": 4166.4, | |
| "valid_targets_min": 1366 | |
| }, | |
| { | |
| "epoch": 3.4030110017371165, | |
| "grad_norm": 0.6424454665235011, | |
| "learning_rate": 2.4435947788994642e-05, | |
| "loss": 0.2441, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11391671001911163, | |
| "step": 2940, | |
| "valid_targets_mean": 3409.9, | |
| "valid_targets_min": 2050 | |
| }, | |
| { | |
| "epoch": 3.4088013896931093, | |
| "grad_norm": 0.6973884546334223, | |
| "learning_rate": 2.4379648955875994e-05, | |
| "loss": 0.2541, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11070425808429718, | |
| "step": 2945, | |
| "valid_targets_mean": 2586.6, | |
| "valid_targets_min": 746 | |
| }, | |
| { | |
| "epoch": 3.4145917776491026, | |
| "grad_norm": 0.6467746768147802, | |
| "learning_rate": 2.4323313647187484e-05, | |
| "loss": 0.2322, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11300863325595856, | |
| "step": 2950, | |
| "valid_targets_mean": 3737.1, | |
| "valid_targets_min": 1538 | |
| }, | |
| { | |
| "epoch": 3.4203821656050954, | |
| "grad_norm": 0.5682308131029035, | |
| "learning_rate": 2.4266942332113387e-05, | |
| "loss": 0.2437, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10985110700130463, | |
| "step": 2955, | |
| "valid_targets_mean": 3841.8, | |
| "valid_targets_min": 2006 | |
| }, | |
| { | |
| "epoch": 3.4261725535610887, | |
| "grad_norm": 0.6742796384715897, | |
| "learning_rate": 2.4210535480137892e-05, | |
| "loss": 0.2566, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12906578183174133, | |
| "step": 2960, | |
| "valid_targets_mean": 3333.8, | |
| "valid_targets_min": 2227 | |
| }, | |
| { | |
| "epoch": 3.4319629415170816, | |
| "grad_norm": 0.5861746371987969, | |
| "learning_rate": 2.415409356104112e-05, | |
| "loss": 0.2517, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11130131781101227, | |
| "step": 2965, | |
| "valid_targets_mean": 4117.2, | |
| "valid_targets_min": 1323 | |
| }, | |
| { | |
| "epoch": 3.437753329473075, | |
| "grad_norm": 0.6457673256720572, | |
| "learning_rate": 2.409761704489526e-05, | |
| "loss": 0.2494, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14976799488067627, | |
| "step": 2970, | |
| "valid_targets_mean": 4786.8, | |
| "valid_targets_min": 2809 | |
| }, | |
| { | |
| "epoch": 3.4435437174290677, | |
| "grad_norm": 0.6343807618423737, | |
| "learning_rate": 2.404110640206064e-05, | |
| "loss": 0.2376, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10772261023521423, | |
| "step": 2975, | |
| "valid_targets_mean": 3456.8, | |
| "valid_targets_min": 1365 | |
| }, | |
| { | |
| "epoch": 3.449334105385061, | |
| "grad_norm": 0.5690937521073623, | |
| "learning_rate": 2.3984562103181803e-05, | |
| "loss": 0.24, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13893179595470428, | |
| "step": 2980, | |
| "valid_targets_mean": 5177.6, | |
| "valid_targets_min": 3669 | |
| }, | |
| { | |
| "epoch": 3.455124493341054, | |
| "grad_norm": 0.6937980857106957, | |
| "learning_rate": 2.3927984619183603e-05, | |
| "loss": 0.226, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11245802044868469, | |
| "step": 2985, | |
| "valid_targets_mean": 2766.8, | |
| "valid_targets_min": 1505 | |
| }, | |
| { | |
| "epoch": 3.4609148812970467, | |
| "grad_norm": 0.658770834743505, | |
| "learning_rate": 2.387137442126726e-05, | |
| "loss": 0.2471, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12539096176624298, | |
| "step": 2990, | |
| "valid_targets_mean": 3500.2, | |
| "valid_targets_min": 2541 | |
| }, | |
| { | |
| "epoch": 3.46670526925304, | |
| "grad_norm": 0.613959744188834, | |
| "learning_rate": 2.3814731980906473e-05, | |
| "loss": 0.2406, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10480717569589615, | |
| "step": 2995, | |
| "valid_targets_mean": 3866.1, | |
| "valid_targets_min": 699 | |
| }, | |
| { | |
| "epoch": 3.472495657209033, | |
| "grad_norm": 0.6400793903784873, | |
| "learning_rate": 2.3758057769843442e-05, | |
| "loss": 0.2386, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11400254815816879, | |
| "step": 3000, | |
| "valid_targets_mean": 3536.1, | |
| "valid_targets_min": 648 | |
| }, | |
| { | |
| "epoch": 3.478286045165026, | |
| "grad_norm": 0.5860990185111562, | |
| "learning_rate": 2.3701352260084985e-05, | |
| "loss": 0.2439, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12419473379850388, | |
| "step": 3005, | |
| "valid_targets_mean": 4126.6, | |
| "valid_targets_min": 2511 | |
| }, | |
| { | |
| "epoch": 3.484076433121019, | |
| "grad_norm": 0.692113026787558, | |
| "learning_rate": 2.3644615923898587e-05, | |
| "loss": 0.2301, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1291700005531311, | |
| "step": 3010, | |
| "valid_targets_mean": 3400.0, | |
| "valid_targets_min": 2831 | |
| }, | |
| { | |
| "epoch": 3.4898668210770123, | |
| "grad_norm": 0.6269413089966849, | |
| "learning_rate": 2.358784923380846e-05, | |
| "loss": 0.242, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11353128403425217, | |
| "step": 3015, | |
| "valid_targets_mean": 3609.0, | |
| "valid_targets_min": 1825 | |
| }, | |
| { | |
| "epoch": 3.495657209033005, | |
| "grad_norm": 0.6882797448449455, | |
| "learning_rate": 2.3531052662591626e-05, | |
| "loss": 0.2333, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13075870275497437, | |
| "step": 3020, | |
| "valid_targets_mean": 3237.6, | |
| "valid_targets_min": 1108 | |
| }, | |
| { | |
| "epoch": 3.5014475969889984, | |
| "grad_norm": 0.6419042933539045, | |
| "learning_rate": 2.347422668327396e-05, | |
| "loss": 0.2337, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08648437261581421, | |
| "step": 3025, | |
| "valid_targets_mean": 2583.8, | |
| "valid_targets_min": 1501 | |
| }, | |
| { | |
| "epoch": 3.5072379849449913, | |
| "grad_norm": 0.6538370467544622, | |
| "learning_rate": 2.3417371769126266e-05, | |
| "loss": 0.2328, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.138927161693573, | |
| "step": 3030, | |
| "valid_targets_mean": 3720.9, | |
| "valid_targets_min": 2261 | |
| }, | |
| { | |
| "epoch": 3.513028372900984, | |
| "grad_norm": 0.6223594532534198, | |
| "learning_rate": 2.3360488393660332e-05, | |
| "loss": 0.2388, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1331544816493988, | |
| "step": 3035, | |
| "valid_targets_mean": 4486.6, | |
| "valid_targets_min": 1716 | |
| }, | |
| { | |
| "epoch": 3.5188187608569774, | |
| "grad_norm": 0.6488372615899336, | |
| "learning_rate": 2.330357703062498e-05, | |
| "loss": 0.2279, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10495131462812424, | |
| "step": 3040, | |
| "valid_targets_mean": 3092.4, | |
| "valid_targets_min": 1281 | |
| }, | |
| { | |
| "epoch": 3.5246091488129707, | |
| "grad_norm": 0.6667477147551848, | |
| "learning_rate": 2.3246638154002122e-05, | |
| "loss": 0.2401, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09248582273721695, | |
| "step": 3045, | |
| "valid_targets_mean": 2772.8, | |
| "valid_targets_min": 1697 | |
| }, | |
| { | |
| "epoch": 3.5303995367689636, | |
| "grad_norm": 0.6109343135643314, | |
| "learning_rate": 2.3189672238002813e-05, | |
| "loss": 0.2405, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10102705657482147, | |
| "step": 3050, | |
| "valid_targets_mean": 4052.0, | |
| "valid_targets_min": 2306 | |
| }, | |
| { | |
| "epoch": 3.5361899247249564, | |
| "grad_norm": 0.5883806614334663, | |
| "learning_rate": 2.3132679757063322e-05, | |
| "loss": 0.2351, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1355743259191513, | |
| "step": 3055, | |
| "valid_targets_mean": 4673.9, | |
| "valid_targets_min": 1965 | |
| }, | |
| { | |
| "epoch": 3.5419803126809497, | |
| "grad_norm": 0.6526251324104081, | |
| "learning_rate": 2.307566118584114e-05, | |
| "loss": 0.233, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09953068941831589, | |
| "step": 3060, | |
| "valid_targets_mean": 2707.0, | |
| "valid_targets_min": 1184 | |
| }, | |
| { | |
| "epoch": 3.5477707006369426, | |
| "grad_norm": 0.6749540950250833, | |
| "learning_rate": 2.3018616999211053e-05, | |
| "loss": 0.2415, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12256621569395065, | |
| "step": 3065, | |
| "valid_targets_mean": 3146.4, | |
| "valid_targets_min": 1401 | |
| }, | |
| { | |
| "epoch": 3.553561088592936, | |
| "grad_norm": 0.6858757618445532, | |
| "learning_rate": 2.296154767226119e-05, | |
| "loss": 0.2383, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1241631805896759, | |
| "step": 3070, | |
| "valid_targets_mean": 3128.6, | |
| "valid_targets_min": 2109 | |
| }, | |
| { | |
| "epoch": 3.5593514765489287, | |
| "grad_norm": 0.6661911939882548, | |
| "learning_rate": 2.2904453680289062e-05, | |
| "loss": 0.2291, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1004391461610794, | |
| "step": 3075, | |
| "valid_targets_mean": 2732.0, | |
| "valid_targets_min": 2061 | |
| }, | |
| { | |
| "epoch": 3.565141864504922, | |
| "grad_norm": 0.6541639028198071, | |
| "learning_rate": 2.2847335498797594e-05, | |
| "loss": 0.2372, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12782573699951172, | |
| "step": 3080, | |
| "valid_targets_mean": 3449.1, | |
| "valid_targets_min": 1580 | |
| }, | |
| { | |
| "epoch": 3.570932252460915, | |
| "grad_norm": 0.6420817772922169, | |
| "learning_rate": 2.279019360349116e-05, | |
| "loss": 0.2356, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09694606065750122, | |
| "step": 3085, | |
| "valid_targets_mean": 2976.4, | |
| "valid_targets_min": 1577 | |
| }, | |
| { | |
| "epoch": 3.576722640416908, | |
| "grad_norm": 0.6115369181788524, | |
| "learning_rate": 2.273302847027166e-05, | |
| "loss": 0.2286, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10978230088949203, | |
| "step": 3090, | |
| "valid_targets_mean": 4427.0, | |
| "valid_targets_min": 2095 | |
| }, | |
| { | |
| "epoch": 3.582513028372901, | |
| "grad_norm": 0.6203970033828395, | |
| "learning_rate": 2.2675840575234505e-05, | |
| "loss": 0.2275, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10813956707715988, | |
| "step": 3095, | |
| "valid_targets_mean": 3615.0, | |
| "valid_targets_min": 1673 | |
| }, | |
| { | |
| "epoch": 3.588303416328894, | |
| "grad_norm": 0.5707467224855606, | |
| "learning_rate": 2.2618630394664688e-05, | |
| "loss": 0.2409, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12412257492542267, | |
| "step": 3100, | |
| "valid_targets_mean": 4451.8, | |
| "valid_targets_min": 2964 | |
| }, | |
| { | |
| "epoch": 3.594093804284887, | |
| "grad_norm": 0.5894831777496278, | |
| "learning_rate": 2.2561398405032807e-05, | |
| "loss": 0.2273, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10445494204759598, | |
| "step": 3105, | |
| "valid_targets_mean": 3968.0, | |
| "valid_targets_min": 2453 | |
| }, | |
| { | |
| "epoch": 3.5998841922408804, | |
| "grad_norm": 0.6194496140271808, | |
| "learning_rate": 2.2504145082991085e-05, | |
| "loss": 0.2392, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11659243702888489, | |
| "step": 3110, | |
| "valid_targets_mean": 3560.6, | |
| "valid_targets_min": 1766 | |
| }, | |
| { | |
| "epoch": 3.6056745801968733, | |
| "grad_norm": 0.6387257036814868, | |
| "learning_rate": 2.2446870905369416e-05, | |
| "loss": 0.2298, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12865771353244781, | |
| "step": 3115, | |
| "valid_targets_mean": 4773.9, | |
| "valid_targets_min": 1757 | |
| }, | |
| { | |
| "epoch": 3.611464968152866, | |
| "grad_norm": 0.6765494323792554, | |
| "learning_rate": 2.2389576349171397e-05, | |
| "loss": 0.2456, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13330453634262085, | |
| "step": 3120, | |
| "valid_targets_mean": 4200.1, | |
| "valid_targets_min": 2203 | |
| }, | |
| { | |
| "epoch": 3.6172553561088594, | |
| "grad_norm": 0.6746232050436143, | |
| "learning_rate": 2.233226189157033e-05, | |
| "loss": 0.2257, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11752544343471527, | |
| "step": 3125, | |
| "valid_targets_mean": 3093.2, | |
| "valid_targets_min": 1955 | |
| }, | |
| { | |
| "epoch": 3.6230457440648522, | |
| "grad_norm": 0.5915526608420583, | |
| "learning_rate": 2.2274928009905267e-05, | |
| "loss": 0.2375, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08885739743709564, | |
| "step": 3130, | |
| "valid_targets_mean": 2910.9, | |
| "valid_targets_min": 1221 | |
| }, | |
| { | |
| "epoch": 3.6288361320208455, | |
| "grad_norm": 0.6390249261029404, | |
| "learning_rate": 2.2217575181677048e-05, | |
| "loss": 0.225, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09216712415218353, | |
| "step": 3135, | |
| "valid_targets_mean": 3756.5, | |
| "valid_targets_min": 2506 | |
| }, | |
| { | |
| "epoch": 3.6346265199768384, | |
| "grad_norm": 0.6412730416327591, | |
| "learning_rate": 2.2160203884544293e-05, | |
| "loss": 0.2209, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10596312582492828, | |
| "step": 3140, | |
| "valid_targets_mean": 4419.8, | |
| "valid_targets_min": 2579 | |
| }, | |
| { | |
| "epoch": 3.6404169079328312, | |
| "grad_norm": 0.6832458547802231, | |
| "learning_rate": 2.2102814596319444e-05, | |
| "loss": 0.2186, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13834300637245178, | |
| "step": 3145, | |
| "valid_targets_mean": 3728.2, | |
| "valid_targets_min": 464 | |
| }, | |
| { | |
| "epoch": 3.6462072958888245, | |
| "grad_norm": 0.621863267559752, | |
| "learning_rate": 2.204540779496477e-05, | |
| "loss": 0.2407, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11769835650920868, | |
| "step": 3150, | |
| "valid_targets_mean": 3660.8, | |
| "valid_targets_min": 1003 | |
| }, | |
| { | |
| "epoch": 3.651997683844818, | |
| "grad_norm": 0.6072219530140257, | |
| "learning_rate": 2.1987983958588413e-05, | |
| "loss": 0.2261, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10092820227146149, | |
| "step": 3155, | |
| "valid_targets_mean": 3318.9, | |
| "valid_targets_min": 1446 | |
| }, | |
| { | |
| "epoch": 3.6577880718008107, | |
| "grad_norm": 0.7761619084531957, | |
| "learning_rate": 2.193054356544039e-05, | |
| "loss": 0.2351, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10936201363801956, | |
| "step": 3160, | |
| "valid_targets_mean": 3807.8, | |
| "valid_targets_min": 1105 | |
| }, | |
| { | |
| "epoch": 3.6635784597568035, | |
| "grad_norm": 0.582808255884101, | |
| "learning_rate": 2.1873087093908588e-05, | |
| "loss": 0.2282, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11235076189041138, | |
| "step": 3165, | |
| "valid_targets_mean": 4257.2, | |
| "valid_targets_min": 1460 | |
| }, | |
| { | |
| "epoch": 3.669368847712797, | |
| "grad_norm": 0.6301435665504801, | |
| "learning_rate": 2.1815615022514826e-05, | |
| "loss": 0.243, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11589612066745758, | |
| "step": 3170, | |
| "valid_targets_mean": 3643.8, | |
| "valid_targets_min": 1172 | |
| }, | |
| { | |
| "epoch": 3.6751592356687897, | |
| "grad_norm": 0.5881241674908854, | |
| "learning_rate": 2.1758127829910835e-05, | |
| "loss": 0.2325, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11802739650011063, | |
| "step": 3175, | |
| "valid_targets_mean": 4325.5, | |
| "valid_targets_min": 1818 | |
| }, | |
| { | |
| "epoch": 3.680949623624783, | |
| "grad_norm": 0.6692753816102861, | |
| "learning_rate": 2.1700625994874276e-05, | |
| "loss": 0.2224, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10609833896160126, | |
| "step": 3180, | |
| "valid_targets_mean": 3680.4, | |
| "valid_targets_min": 1535 | |
| }, | |
| { | |
| "epoch": 3.686740011580776, | |
| "grad_norm": 0.6327977443669413, | |
| "learning_rate": 2.1643109996304768e-05, | |
| "loss": 0.2223, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11642224341630936, | |
| "step": 3185, | |
| "valid_targets_mean": 4250.5, | |
| "valid_targets_min": 1814 | |
| }, | |
| { | |
| "epoch": 3.692530399536769, | |
| "grad_norm": 0.7185575308202862, | |
| "learning_rate": 2.158558031321988e-05, | |
| "loss": 0.2197, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10359126329421997, | |
| "step": 3190, | |
| "valid_targets_mean": 2750.5, | |
| "valid_targets_min": 1073 | |
| }, | |
| { | |
| "epoch": 3.698320787492762, | |
| "grad_norm": 0.6112249812545777, | |
| "learning_rate": 2.1528037424751176e-05, | |
| "loss": 0.2214, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11483891308307648, | |
| "step": 3195, | |
| "valid_targets_mean": 4409.8, | |
| "valid_targets_min": 3131 | |
| }, | |
| { | |
| "epoch": 3.7041111754487552, | |
| "grad_norm": 0.622363696526355, | |
| "learning_rate": 2.1470481810140168e-05, | |
| "loss": 0.2146, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09661306440830231, | |
| "step": 3200, | |
| "valid_targets_mean": 3341.8, | |
| "valid_targets_min": 1079 | |
| }, | |
| { | |
| "epoch": 3.709901563404748, | |
| "grad_norm": 0.6207658574507436, | |
| "learning_rate": 2.1412913948734365e-05, | |
| "loss": 0.2099, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10788443684577942, | |
| "step": 3205, | |
| "valid_targets_mean": 3970.4, | |
| "valid_targets_min": 1143 | |
| }, | |
| { | |
| "epoch": 3.715691951360741, | |
| "grad_norm": 0.6145454281573289, | |
| "learning_rate": 2.135533431998329e-05, | |
| "loss": 0.2204, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09064505994319916, | |
| "step": 3210, | |
| "valid_targets_mean": 3327.1, | |
| "valid_targets_min": 1981 | |
| }, | |
| { | |
| "epoch": 3.7214823393167342, | |
| "grad_norm": 0.7142208351524535, | |
| "learning_rate": 2.129774340343446e-05, | |
| "loss": 0.248, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10260207951068878, | |
| "step": 3215, | |
| "valid_targets_mean": 2796.6, | |
| "valid_targets_min": 1410 | |
| }, | |
| { | |
| "epoch": 3.7272727272727275, | |
| "grad_norm": 0.6313431287384368, | |
| "learning_rate": 2.1240141678729387e-05, | |
| "loss": 0.222, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09830041974782944, | |
| "step": 3220, | |
| "valid_targets_mean": 2989.8, | |
| "valid_targets_min": 1691 | |
| }, | |
| { | |
| "epoch": 3.7330631152287204, | |
| "grad_norm": 0.9358403011904364, | |
| "learning_rate": 2.1182529625599617e-05, | |
| "loss": 0.2318, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.124494269490242, | |
| "step": 3225, | |
| "valid_targets_mean": 4091.6, | |
| "valid_targets_min": 2318 | |
| }, | |
| { | |
| "epoch": 3.738853503184713, | |
| "grad_norm": 0.5973542139833429, | |
| "learning_rate": 2.1124907723862706e-05, | |
| "loss": 0.2204, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09477721154689789, | |
| "step": 3230, | |
| "valid_targets_mean": 3572.8, | |
| "valid_targets_min": 1509 | |
| }, | |
| { | |
| "epoch": 3.7446438911407065, | |
| "grad_norm": 0.6398607876411986, | |
| "learning_rate": 2.106727645341824e-05, | |
| "loss": 0.2263, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11783312261104584, | |
| "step": 3235, | |
| "valid_targets_mean": 3824.9, | |
| "valid_targets_min": 1601 | |
| }, | |
| { | |
| "epoch": 3.7504342790966994, | |
| "grad_norm": 0.6817149155548572, | |
| "learning_rate": 2.100963629424382e-05, | |
| "loss": 0.2226, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1156693622469902, | |
| "step": 3240, | |
| "valid_targets_mean": 3558.9, | |
| "valid_targets_min": 1281 | |
| }, | |
| { | |
| "epoch": 3.7562246670526926, | |
| "grad_norm": 0.5728050591821362, | |
| "learning_rate": 2.095198772639109e-05, | |
| "loss": 0.222, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09249010682106018, | |
| "step": 3245, | |
| "valid_targets_mean": 3256.1, | |
| "valid_targets_min": 1445 | |
| }, | |
| { | |
| "epoch": 3.7620150550086855, | |
| "grad_norm": 0.66348305148334, | |
| "learning_rate": 2.089433122998172e-05, | |
| "loss": 0.2269, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13942071795463562, | |
| "step": 3250, | |
| "valid_targets_mean": 4249.1, | |
| "valid_targets_min": 3677 | |
| }, | |
| { | |
| "epoch": 3.767805442964679, | |
| "grad_norm": 0.650179732240423, | |
| "learning_rate": 2.0836667285203403e-05, | |
| "loss": 0.2216, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.132462739944458, | |
| "step": 3255, | |
| "valid_targets_mean": 4770.8, | |
| "valid_targets_min": 2121 | |
| }, | |
| { | |
| "epoch": 3.7735958309206716, | |
| "grad_norm": 0.608692207261934, | |
| "learning_rate": 2.077899637230588e-05, | |
| "loss": 0.2154, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11029618978500366, | |
| "step": 3260, | |
| "valid_targets_mean": 3527.8, | |
| "valid_targets_min": 1202 | |
| }, | |
| { | |
| "epoch": 3.779386218876665, | |
| "grad_norm": 0.6256858394345893, | |
| "learning_rate": 2.0721318971596915e-05, | |
| "loss": 0.238, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11849434673786163, | |
| "step": 3265, | |
| "valid_targets_mean": 4491.6, | |
| "valid_targets_min": 3504 | |
| }, | |
| { | |
| "epoch": 3.7851766068326578, | |
| "grad_norm": 0.574664428323433, | |
| "learning_rate": 2.0663635563438306e-05, | |
| "loss": 0.2344, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1048889011144638, | |
| "step": 3270, | |
| "valid_targets_mean": 4129.0, | |
| "valid_targets_min": 1849 | |
| }, | |
| { | |
| "epoch": 3.7909669947886506, | |
| "grad_norm": 0.7801975367950815, | |
| "learning_rate": 2.0605946628241895e-05, | |
| "loss": 0.2395, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16848134994506836, | |
| "step": 3275, | |
| "valid_targets_mean": 4321.4, | |
| "valid_targets_min": 3034 | |
| }, | |
| { | |
| "epoch": 3.796757382744644, | |
| "grad_norm": 0.6443689938605106, | |
| "learning_rate": 2.0548252646465544e-05, | |
| "loss": 0.2319, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12259537726640701, | |
| "step": 3280, | |
| "valid_targets_mean": 4204.4, | |
| "valid_targets_min": 2310 | |
| }, | |
| { | |
| "epoch": 3.802547770700637, | |
| "grad_norm": 0.6629375966318463, | |
| "learning_rate": 2.0490554098609144e-05, | |
| "loss": 0.2357, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11784237623214722, | |
| "step": 3285, | |
| "valid_targets_mean": 4133.1, | |
| "valid_targets_min": 2310 | |
| }, | |
| { | |
| "epoch": 3.80833815865663, | |
| "grad_norm": 0.6779347834232181, | |
| "learning_rate": 2.0432851465210618e-05, | |
| "loss": 0.2402, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1057584285736084, | |
| "step": 3290, | |
| "valid_targets_mean": 2686.6, | |
| "valid_targets_min": 802 | |
| }, | |
| { | |
| "epoch": 3.814128546612623, | |
| "grad_norm": 0.6612416426138654, | |
| "learning_rate": 2.0375145226841916e-05, | |
| "loss": 0.2425, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1279018074274063, | |
| "step": 3295, | |
| "valid_targets_mean": 3936.5, | |
| "valid_targets_min": 1433 | |
| }, | |
| { | |
| "epoch": 3.819918934568616, | |
| "grad_norm": 0.5944207402613865, | |
| "learning_rate": 2.0317435864105017e-05, | |
| "loss": 0.2222, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12686294317245483, | |
| "step": 3300, | |
| "valid_targets_mean": 4604.6, | |
| "valid_targets_min": 2467 | |
| }, | |
| { | |
| "epoch": 3.825709322524609, | |
| "grad_norm": 0.8029093931170834, | |
| "learning_rate": 2.025972385762791e-05, | |
| "loss": 0.234, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10743986815214157, | |
| "step": 3305, | |
| "valid_targets_mean": 3055.6, | |
| "valid_targets_min": 1501 | |
| }, | |
| { | |
| "epoch": 3.8314997104806023, | |
| "grad_norm": 0.7435607132336812, | |
| "learning_rate": 2.0202009688060603e-05, | |
| "loss": 0.2365, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.140691339969635, | |
| "step": 3310, | |
| "valid_targets_mean": 3669.0, | |
| "valid_targets_min": 2260 | |
| }, | |
| { | |
| "epoch": 3.837290098436595, | |
| "grad_norm": 0.6617232566726545, | |
| "learning_rate": 2.0144293836071132e-05, | |
| "loss": 0.2304, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11230485141277313, | |
| "step": 3315, | |
| "valid_targets_mean": 3529.6, | |
| "valid_targets_min": 1325 | |
| }, | |
| { | |
| "epoch": 3.843080486392588, | |
| "grad_norm": 0.5750730658739346, | |
| "learning_rate": 2.0086576782341537e-05, | |
| "loss": 0.2272, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1011202335357666, | |
| "step": 3320, | |
| "valid_targets_mean": 4397.6, | |
| "valid_targets_min": 2754 | |
| }, | |
| { | |
| "epoch": 3.8488708743485813, | |
| "grad_norm": 0.6146915424507076, | |
| "learning_rate": 2.0028859007563857e-05, | |
| "loss": 0.2164, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09985917806625366, | |
| "step": 3325, | |
| "valid_targets_mean": 3566.8, | |
| "valid_targets_min": 1991 | |
| }, | |
| { | |
| "epoch": 3.8546612623045746, | |
| "grad_norm": 0.6660943821357719, | |
| "learning_rate": 1.997114099243615e-05, | |
| "loss": 0.2326, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12243643403053284, | |
| "step": 3330, | |
| "valid_targets_mean": 2909.9, | |
| "valid_targets_min": 1076 | |
| }, | |
| { | |
| "epoch": 3.8604516502605675, | |
| "grad_norm": 0.6308468996973874, | |
| "learning_rate": 1.9913423217658466e-05, | |
| "loss": 0.2161, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08638469874858856, | |
| "step": 3335, | |
| "valid_targets_mean": 3275.1, | |
| "valid_targets_min": 521 | |
| }, | |
| { | |
| "epoch": 3.8662420382165603, | |
| "grad_norm": 0.5950358153672971, | |
| "learning_rate": 1.9855706163928868e-05, | |
| "loss": 0.2383, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09733293950557709, | |
| "step": 3340, | |
| "valid_targets_mean": 3767.1, | |
| "valid_targets_min": 2185 | |
| }, | |
| { | |
| "epoch": 3.8720324261725536, | |
| "grad_norm": 0.6053995285078753, | |
| "learning_rate": 1.9797990311939394e-05, | |
| "loss": 0.2311, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1037420853972435, | |
| "step": 3345, | |
| "valid_targets_mean": 3945.8, | |
| "valid_targets_min": 2024 | |
| }, | |
| { | |
| "epoch": 3.8778228141285465, | |
| "grad_norm": 0.6214312013380663, | |
| "learning_rate": 1.97402761423721e-05, | |
| "loss": 0.2304, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0892077088356018, | |
| "step": 3350, | |
| "valid_targets_mean": 3204.5, | |
| "valid_targets_min": 1777 | |
| }, | |
| { | |
| "epoch": 3.8836132020845398, | |
| "grad_norm": 0.643149933538212, | |
| "learning_rate": 1.968256413589499e-05, | |
| "loss": 0.222, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09115341305732727, | |
| "step": 3355, | |
| "valid_targets_mean": 3268.5, | |
| "valid_targets_min": 1965 | |
| }, | |
| { | |
| "epoch": 3.8894035900405326, | |
| "grad_norm": 0.6291381948112222, | |
| "learning_rate": 1.962485477315809e-05, | |
| "loss": 0.2364, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13257244229316711, | |
| "step": 3360, | |
| "valid_targets_mean": 4630.0, | |
| "valid_targets_min": 1307 | |
| }, | |
| { | |
| "epoch": 3.895193977996526, | |
| "grad_norm": 0.6454990402044124, | |
| "learning_rate": 1.956714853478939e-05, | |
| "loss": 0.2297, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1468641608953476, | |
| "step": 3365, | |
| "valid_targets_mean": 5090.8, | |
| "valid_targets_min": 2762 | |
| }, | |
| { | |
| "epoch": 3.9009843659525187, | |
| "grad_norm": 0.7135917015837402, | |
| "learning_rate": 1.9509445901390863e-05, | |
| "loss": 0.2206, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10307130217552185, | |
| "step": 3370, | |
| "valid_targets_mean": 2930.9, | |
| "valid_targets_min": 1035 | |
| }, | |
| { | |
| "epoch": 3.906774753908512, | |
| "grad_norm": 0.6143842180197991, | |
| "learning_rate": 1.9451747353534463e-05, | |
| "loss": 0.2173, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13558322191238403, | |
| "step": 3375, | |
| "valid_targets_mean": 4369.2, | |
| "valid_targets_min": 1876 | |
| }, | |
| { | |
| "epoch": 3.912565141864505, | |
| "grad_norm": 0.5959476877785364, | |
| "learning_rate": 1.9394053371758108e-05, | |
| "loss": 0.2209, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09872700273990631, | |
| "step": 3380, | |
| "valid_targets_mean": 3840.5, | |
| "valid_targets_min": 2303 | |
| }, | |
| { | |
| "epoch": 3.9183555298204977, | |
| "grad_norm": 0.6416421146123354, | |
| "learning_rate": 1.93363644365617e-05, | |
| "loss": 0.219, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09142628312110901, | |
| "step": 3385, | |
| "valid_targets_mean": 3454.0, | |
| "valid_targets_min": 1434 | |
| }, | |
| { | |
| "epoch": 3.924145917776491, | |
| "grad_norm": 0.5691149807130056, | |
| "learning_rate": 1.9278681028403095e-05, | |
| "loss": 0.2182, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1255941390991211, | |
| "step": 3390, | |
| "valid_targets_mean": 4752.5, | |
| "valid_targets_min": 1496 | |
| }, | |
| { | |
| "epoch": 3.9299363057324843, | |
| "grad_norm": 0.6745112014980869, | |
| "learning_rate": 1.9221003627694127e-05, | |
| "loss": 0.2151, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1109854057431221, | |
| "step": 3395, | |
| "valid_targets_mean": 3358.0, | |
| "valid_targets_min": 1518 | |
| }, | |
| { | |
| "epoch": 3.935726693688477, | |
| "grad_norm": 0.6071462519506634, | |
| "learning_rate": 1.9163332714796604e-05, | |
| "loss": 0.2233, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10561128705739975, | |
| "step": 3400, | |
| "valid_targets_mean": 4146.1, | |
| "valid_targets_min": 1152 | |
| }, | |
| { | |
| "epoch": 3.94151708164447, | |
| "grad_norm": 0.6426584726925665, | |
| "learning_rate": 1.9105668770018287e-05, | |
| "loss": 0.2312, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11722976714372635, | |
| "step": 3405, | |
| "valid_targets_mean": 3758.4, | |
| "valid_targets_min": 1598 | |
| }, | |
| { | |
| "epoch": 3.9473074696004633, | |
| "grad_norm": 0.617643112012234, | |
| "learning_rate": 1.9048012273608912e-05, | |
| "loss": 0.2422, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1155039593577385, | |
| "step": 3410, | |
| "valid_targets_mean": 4272.2, | |
| "valid_targets_min": 1611 | |
| }, | |
| { | |
| "epoch": 3.953097857556456, | |
| "grad_norm": 0.6750768602220908, | |
| "learning_rate": 1.8990363705756182e-05, | |
| "loss": 0.2171, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09991435706615448, | |
| "step": 3415, | |
| "valid_targets_mean": 3249.9, | |
| "valid_targets_min": 1490 | |
| }, | |
| { | |
| "epoch": 3.9588882455124494, | |
| "grad_norm": 0.6683691782809446, | |
| "learning_rate": 1.8932723546581767e-05, | |
| "loss": 0.2207, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09730266779661179, | |
| "step": 3420, | |
| "valid_targets_mean": 2868.1, | |
| "valid_targets_min": 2355 | |
| }, | |
| { | |
| "epoch": 3.9646786334684423, | |
| "grad_norm": 0.6833310701250812, | |
| "learning_rate": 1.88750922761373e-05, | |
| "loss": 0.2236, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11862599104642868, | |
| "step": 3425, | |
| "valid_targets_mean": 3776.5, | |
| "valid_targets_min": 1743 | |
| }, | |
| { | |
| "epoch": 3.9704690214244356, | |
| "grad_norm": 0.555511474925617, | |
| "learning_rate": 1.881747037440039e-05, | |
| "loss": 0.2212, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09805724769830704, | |
| "step": 3430, | |
| "valid_targets_mean": 4739.0, | |
| "valid_targets_min": 2933 | |
| }, | |
| { | |
| "epoch": 3.9762594093804284, | |
| "grad_norm": 0.6519285293424276, | |
| "learning_rate": 1.8759858321270616e-05, | |
| "loss": 0.2305, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12541580200195312, | |
| "step": 3435, | |
| "valid_targets_mean": 3573.1, | |
| "valid_targets_min": 2433 | |
| }, | |
| { | |
| "epoch": 3.9820497973364217, | |
| "grad_norm": 0.6896268921953539, | |
| "learning_rate": 1.8702256596565547e-05, | |
| "loss": 0.2301, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12650898098945618, | |
| "step": 3440, | |
| "valid_targets_mean": 3635.6, | |
| "valid_targets_min": 2100 | |
| }, | |
| { | |
| "epoch": 3.9878401852924146, | |
| "grad_norm": 0.7276116433866298, | |
| "learning_rate": 1.864466568001671e-05, | |
| "loss": 0.2252, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09770257771015167, | |
| "step": 3445, | |
| "valid_targets_mean": 3246.2, | |
| "valid_targets_min": 1365 | |
| }, | |
| { | |
| "epoch": 3.9936305732484074, | |
| "grad_norm": 0.7005066198558177, | |
| "learning_rate": 1.858708605126563e-05, | |
| "loss": 0.2269, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1187184602022171, | |
| "step": 3450, | |
| "valid_targets_mean": 4794.2, | |
| "valid_targets_min": 1572 | |
| }, | |
| { | |
| "epoch": 3.9994209612044007, | |
| "grad_norm": 1.1531137671857628, | |
| "learning_rate": 1.8529518189859842e-05, | |
| "loss": 0.2304, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11856088787317276, | |
| "step": 3455, | |
| "valid_targets_mean": 3688.5, | |
| "valid_targets_min": 876 | |
| }, | |
| { | |
| "epoch": 4.004632310364794, | |
| "grad_norm": 0.6338487908223315, | |
| "learning_rate": 1.847196257524883e-05, | |
| "loss": 0.1867, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1476893126964569, | |
| "step": 3460, | |
| "valid_targets_mean": 8029.9, | |
| "valid_targets_min": 6142 | |
| }, | |
| { | |
| "epoch": 4.010422698320787, | |
| "grad_norm": 0.47487343877733607, | |
| "learning_rate": 1.8414419686780124e-05, | |
| "loss": 0.1912, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09128034114837646, | |
| "step": 3465, | |
| "valid_targets_mean": 6664.0, | |
| "valid_targets_min": 5470 | |
| }, | |
| { | |
| "epoch": 4.016213086276781, | |
| "grad_norm": 0.4441691540656432, | |
| "learning_rate": 1.8356890003695242e-05, | |
| "loss": 0.1825, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09573540836572647, | |
| "step": 3470, | |
| "valid_targets_mean": 7126.9, | |
| "valid_targets_min": 5368 | |
| }, | |
| { | |
| "epoch": 4.022003474232774, | |
| "grad_norm": 0.6934913878421209, | |
| "learning_rate": 1.829937400512573e-05, | |
| "loss": 0.1769, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08647888153791428, | |
| "step": 3475, | |
| "valid_targets_mean": 7634.9, | |
| "valid_targets_min": 5614 | |
| }, | |
| { | |
| "epoch": 4.027793862188767, | |
| "grad_norm": 0.3845706850810253, | |
| "learning_rate": 1.824187217008917e-05, | |
| "loss": 0.1828, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08422435820102692, | |
| "step": 3480, | |
| "valid_targets_mean": 8630.8, | |
| "valid_targets_min": 6362 | |
| }, | |
| { | |
| "epoch": 4.033584250144759, | |
| "grad_norm": 0.4508861479189168, | |
| "learning_rate": 1.8184384977485177e-05, | |
| "loss": 0.1859, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09215463697910309, | |
| "step": 3485, | |
| "valid_targets_mean": 7383.6, | |
| "valid_targets_min": 5844 | |
| }, | |
| { | |
| "epoch": 4.039374638100753, | |
| "grad_norm": 0.40329975694327114, | |
| "learning_rate": 1.812691290609142e-05, | |
| "loss": 0.1963, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10451547056436539, | |
| "step": 3490, | |
| "valid_targets_mean": 8017.0, | |
| "valid_targets_min": 5283 | |
| }, | |
| { | |
| "epoch": 4.045165026056746, | |
| "grad_norm": 0.4056281920802866, | |
| "learning_rate": 1.8069456434559618e-05, | |
| "loss": 0.1816, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07850828766822815, | |
| "step": 3495, | |
| "valid_targets_mean": 7051.1, | |
| "valid_targets_min": 4135 | |
| }, | |
| { | |
| "epoch": 4.050955414012739, | |
| "grad_norm": 0.4265931631056118, | |
| "learning_rate": 1.8012016041411593e-05, | |
| "loss": 0.1775, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07563488185405731, | |
| "step": 3500, | |
| "valid_targets_mean": 5370.4, | |
| "valid_targets_min": 3919 | |
| }, | |
| { | |
| "epoch": 4.056745801968732, | |
| "grad_norm": 0.45089862246410817, | |
| "learning_rate": 1.7954592205035235e-05, | |
| "loss": 0.1629, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07701076567173004, | |
| "step": 3505, | |
| "valid_targets_mean": 6063.6, | |
| "valid_targets_min": 4863 | |
| }, | |
| { | |
| "epoch": 4.0625361899247245, | |
| "grad_norm": 0.43651516381480626, | |
| "learning_rate": 1.7897185403680562e-05, | |
| "loss": 0.181, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08604348450899124, | |
| "step": 3510, | |
| "valid_targets_mean": 7001.2, | |
| "valid_targets_min": 6020 | |
| }, | |
| { | |
| "epoch": 4.068326577880718, | |
| "grad_norm": 0.44093517531962334, | |
| "learning_rate": 1.7839796115455707e-05, | |
| "loss": 0.1716, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.085826575756073, | |
| "step": 3515, | |
| "valid_targets_mean": 6181.9, | |
| "valid_targets_min": 4669 | |
| }, | |
| { | |
| "epoch": 4.074116965836711, | |
| "grad_norm": 0.3999835630513633, | |
| "learning_rate": 1.7782424818322955e-05, | |
| "loss": 0.1774, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07994566857814789, | |
| "step": 3520, | |
| "valid_targets_mean": 6250.2, | |
| "valid_targets_min": 4900 | |
| }, | |
| { | |
| "epoch": 4.079907353792704, | |
| "grad_norm": 0.40031048525427254, | |
| "learning_rate": 1.7725071990094743e-05, | |
| "loss": 0.17, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07790108025074005, | |
| "step": 3525, | |
| "valid_targets_mean": 7156.4, | |
| "valid_targets_min": 5323 | |
| }, | |
| { | |
| "epoch": 4.085697741748697, | |
| "grad_norm": 0.4184529184232665, | |
| "learning_rate": 1.766773810842968e-05, | |
| "loss": 0.1731, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08983001112937927, | |
| "step": 3530, | |
| "valid_targets_mean": 7215.8, | |
| "valid_targets_min": 4720 | |
| }, | |
| { | |
| "epoch": 4.091488129704691, | |
| "grad_norm": 0.416428854975788, | |
| "learning_rate": 1.761042365082861e-05, | |
| "loss": 0.1788, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08676470816135406, | |
| "step": 3535, | |
| "valid_targets_mean": 6550.8, | |
| "valid_targets_min": 5739 | |
| }, | |
| { | |
| "epoch": 4.097278517660683, | |
| "grad_norm": 0.4216208509272915, | |
| "learning_rate": 1.755312909463059e-05, | |
| "loss": 0.1717, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08913668990135193, | |
| "step": 3540, | |
| "valid_targets_mean": 6726.8, | |
| "valid_targets_min": 4540 | |
| }, | |
| { | |
| "epoch": 4.103068905616676, | |
| "grad_norm": 0.5196781964099998, | |
| "learning_rate": 1.749585491700892e-05, | |
| "loss": 0.1789, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09615586698055267, | |
| "step": 3545, | |
| "valid_targets_mean": 4573.4, | |
| "valid_targets_min": 1039 | |
| }, | |
| { | |
| "epoch": 4.108859293572669, | |
| "grad_norm": 0.43286538062723917, | |
| "learning_rate": 1.7438601594967196e-05, | |
| "loss": 0.189, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08920808136463165, | |
| "step": 3550, | |
| "valid_targets_mean": 6430.0, | |
| "valid_targets_min": 5412 | |
| }, | |
| { | |
| "epoch": 4.114649681528663, | |
| "grad_norm": 0.7025560718319794, | |
| "learning_rate": 1.7381369605335312e-05, | |
| "loss": 0.1692, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.02563580311834812, | |
| "step": 3555, | |
| "valid_targets_mean": 583.1, | |
| "valid_targets_min": 158 | |
| }, | |
| { | |
| "epoch": 4.120440069484656, | |
| "grad_norm": 6.598428220612047, | |
| "learning_rate": 1.7324159424765502e-05, | |
| "loss": 0.1612, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08991201221942902, | |
| "step": 3560, | |
| "valid_targets_mean": 8230.9, | |
| "valid_targets_min": 5061 | |
| }, | |
| { | |
| "epoch": 4.1262304574406485, | |
| "grad_norm": 0.44186789396424986, | |
| "learning_rate": 1.726697152972835e-05, | |
| "loss": 0.1743, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09638067334890366, | |
| "step": 3565, | |
| "valid_targets_mean": 7238.9, | |
| "valid_targets_min": 5112 | |
| }, | |
| { | |
| "epoch": 4.132020845396641, | |
| "grad_norm": 0.4450781677124384, | |
| "learning_rate": 1.7209806396508847e-05, | |
| "loss": 0.1726, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07829023897647858, | |
| "step": 3570, | |
| "valid_targets_mean": 6322.8, | |
| "valid_targets_min": 4867 | |
| }, | |
| { | |
| "epoch": 4.137811233352634, | |
| "grad_norm": 0.4122268360572617, | |
| "learning_rate": 1.7152664501202413e-05, | |
| "loss": 0.1795, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08773788809776306, | |
| "step": 3575, | |
| "valid_targets_mean": 7887.0, | |
| "valid_targets_min": 6004 | |
| }, | |
| { | |
| "epoch": 4.143601621308628, | |
| "grad_norm": 0.39107096983030126, | |
| "learning_rate": 1.7095546319710944e-05, | |
| "loss": 0.1698, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07779907435178757, | |
| "step": 3580, | |
| "valid_targets_mean": 7090.2, | |
| "valid_targets_min": 4917 | |
| }, | |
| { | |
| "epoch": 4.149392009264621, | |
| "grad_norm": 0.41031771658626737, | |
| "learning_rate": 1.703845232773881e-05, | |
| "loss": 0.1651, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08271464705467224, | |
| "step": 3585, | |
| "valid_targets_mean": 7441.9, | |
| "valid_targets_min": 4202 | |
| }, | |
| { | |
| "epoch": 4.155182397220614, | |
| "grad_norm": 0.5269944341101026, | |
| "learning_rate": 1.698138300078895e-05, | |
| "loss": 0.1516, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08067260682582855, | |
| "step": 3590, | |
| "valid_targets_mean": 7901.8, | |
| "valid_targets_min": 5060 | |
| }, | |
| { | |
| "epoch": 4.1609727851766065, | |
| "grad_norm": 0.40568727713688824, | |
| "learning_rate": 1.692433881415887e-05, | |
| "loss": 0.1538, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07333912700414658, | |
| "step": 3595, | |
| "valid_targets_mean": 8557.6, | |
| "valid_targets_min": 6331 | |
| }, | |
| { | |
| "epoch": 4.1667631731326, | |
| "grad_norm": 0.4594855492171708, | |
| "learning_rate": 1.686732024293668e-05, | |
| "loss": 0.1793, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08284049481153488, | |
| "step": 3600, | |
| "valid_targets_mean": 4976.4, | |
| "valid_targets_min": 2847 | |
| }, | |
| { | |
| "epoch": 4.172553561088593, | |
| "grad_norm": 0.4848900085631287, | |
| "learning_rate": 1.6810327761997193e-05, | |
| "loss": 0.1611, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09487798810005188, | |
| "step": 3605, | |
| "valid_targets_mean": 7129.1, | |
| "valid_targets_min": 5331 | |
| }, | |
| { | |
| "epoch": 4.178343949044586, | |
| "grad_norm": 0.45541337635011625, | |
| "learning_rate": 1.6753361845997888e-05, | |
| "loss": 0.182, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09323504567146301, | |
| "step": 3610, | |
| "valid_targets_mean": 6864.0, | |
| "valid_targets_min": 4697 | |
| }, | |
| { | |
| "epoch": 4.184134337000579, | |
| "grad_norm": 0.48558524786208335, | |
| "learning_rate": 1.6696422969375027e-05, | |
| "loss": 0.1892, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08608924597501755, | |
| "step": 3615, | |
| "valid_targets_mean": 6631.9, | |
| "valid_targets_min": 4623 | |
| }, | |
| { | |
| "epoch": 4.1899247249565725, | |
| "grad_norm": 0.4337802573922538, | |
| "learning_rate": 1.6639511606339674e-05, | |
| "loss": 0.181, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08183959126472473, | |
| "step": 3620, | |
| "valid_targets_mean": 6330.4, | |
| "valid_targets_min": 5424 | |
| }, | |
| { | |
| "epoch": 4.195715112912565, | |
| "grad_norm": 0.4401264336839986, | |
| "learning_rate": 1.6582628230873737e-05, | |
| "loss": 0.1753, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08481311798095703, | |
| "step": 3625, | |
| "valid_targets_mean": 6814.8, | |
| "valid_targets_min": 4964 | |
| }, | |
| { | |
| "epoch": 4.201505500868558, | |
| "grad_norm": 0.46448901960485245, | |
| "learning_rate": 1.652577331672605e-05, | |
| "loss": 0.1816, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09653014689683914, | |
| "step": 3630, | |
| "valid_targets_mean": 6915.8, | |
| "valid_targets_min": 5321 | |
| }, | |
| { | |
| "epoch": 4.207295888824551, | |
| "grad_norm": 0.43792282662310117, | |
| "learning_rate": 1.646894733740838e-05, | |
| "loss": 0.1749, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09070654213428497, | |
| "step": 3635, | |
| "valid_targets_mean": 6464.5, | |
| "valid_targets_min": 5185 | |
| }, | |
| { | |
| "epoch": 4.213086276780544, | |
| "grad_norm": 0.42338152090227527, | |
| "learning_rate": 1.6412150766191545e-05, | |
| "loss": 0.1772, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09077668935060501, | |
| "step": 3640, | |
| "valid_targets_mean": 7048.6, | |
| "valid_targets_min": 4913 | |
| }, | |
| { | |
| "epoch": 4.218876664736538, | |
| "grad_norm": 0.4228905286663746, | |
| "learning_rate": 1.635538407610142e-05, | |
| "loss": 0.1808, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09877828508615494, | |
| "step": 3645, | |
| "valid_targets_mean": 7469.2, | |
| "valid_targets_min": 5426 | |
| }, | |
| { | |
| "epoch": 4.2246670526925305, | |
| "grad_norm": 0.4518896532340265, | |
| "learning_rate": 1.6298647739915018e-05, | |
| "loss": 0.1795, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09248585999011993, | |
| "step": 3650, | |
| "valid_targets_mean": 6819.4, | |
| "valid_targets_min": 5286 | |
| }, | |
| { | |
| "epoch": 4.230457440648523, | |
| "grad_norm": 0.43493724361595565, | |
| "learning_rate": 1.624194223015656e-05, | |
| "loss": 0.1786, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09276282787322998, | |
| "step": 3655, | |
| "valid_targets_mean": 7107.5, | |
| "valid_targets_min": 5201 | |
| }, | |
| { | |
| "epoch": 4.236247828604516, | |
| "grad_norm": 0.6009769366191408, | |
| "learning_rate": 1.6185268019093534e-05, | |
| "loss": 0.1757, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08548557013273239, | |
| "step": 3660, | |
| "valid_targets_mean": 5642.6, | |
| "valid_targets_min": 4838 | |
| }, | |
| { | |
| "epoch": 4.24203821656051, | |
| "grad_norm": 0.9590245633099599, | |
| "learning_rate": 1.612862557873275e-05, | |
| "loss": 0.1447, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14384236931800842, | |
| "step": 3665, | |
| "valid_targets_mean": 2054.1, | |
| "valid_targets_min": 134 | |
| }, | |
| { | |
| "epoch": 4.247828604516503, | |
| "grad_norm": 0.48201038621071807, | |
| "learning_rate": 1.6072015380816407e-05, | |
| "loss": 0.1799, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08267612755298615, | |
| "step": 3670, | |
| "valid_targets_mean": 6686.4, | |
| "valid_targets_min": 5256 | |
| }, | |
| { | |
| "epoch": 4.253618992472496, | |
| "grad_norm": 0.44266137339887035, | |
| "learning_rate": 1.6015437896818204e-05, | |
| "loss": 0.1718, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0936516746878624, | |
| "step": 3675, | |
| "valid_targets_mean": 6216.6, | |
| "valid_targets_min": 5542 | |
| }, | |
| { | |
| "epoch": 4.2594093804284885, | |
| "grad_norm": 0.4319241530250814, | |
| "learning_rate": 1.5958893597939363e-05, | |
| "loss": 0.1768, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08643292635679245, | |
| "step": 3680, | |
| "valid_targets_mean": 6486.1, | |
| "valid_targets_min": 5531 | |
| }, | |
| { | |
| "epoch": 4.265199768384482, | |
| "grad_norm": 0.44892852652887477, | |
| "learning_rate": 1.5902382955104745e-05, | |
| "loss": 0.1713, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08577564358711243, | |
| "step": 3685, | |
| "valid_targets_mean": 5797.4, | |
| "valid_targets_min": 3911 | |
| }, | |
| { | |
| "epoch": 4.270990156340475, | |
| "grad_norm": 0.45772014152728996, | |
| "learning_rate": 1.5845906438958884e-05, | |
| "loss": 0.1733, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08880448341369629, | |
| "step": 3690, | |
| "valid_targets_mean": 6357.5, | |
| "valid_targets_min": 5195 | |
| }, | |
| { | |
| "epoch": 4.276780544296468, | |
| "grad_norm": 0.4873851162289285, | |
| "learning_rate": 1.5789464519862108e-05, | |
| "loss": 0.1785, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0878804624080658, | |
| "step": 3695, | |
| "valid_targets_mean": 5808.9, | |
| "valid_targets_min": 4959 | |
| }, | |
| { | |
| "epoch": 4.282570932252461, | |
| "grad_norm": 0.5766157379649499, | |
| "learning_rate": 1.5733057667886617e-05, | |
| "loss": 0.1734, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.056061677634716034, | |
| "step": 3700, | |
| "valid_targets_mean": 2591.5, | |
| "valid_targets_min": 842 | |
| }, | |
| { | |
| "epoch": 4.288361320208454, | |
| "grad_norm": 0.4927049290514214, | |
| "learning_rate": 1.567668635281253e-05, | |
| "loss": 0.1449, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08399844169616699, | |
| "step": 3705, | |
| "valid_targets_mean": 6006.9, | |
| "valid_targets_min": 4769 | |
| }, | |
| { | |
| "epoch": 4.294151708164447, | |
| "grad_norm": 0.5355797089915041, | |
| "learning_rate": 1.5620351044124013e-05, | |
| "loss": 0.1723, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0904797911643982, | |
| "step": 3710, | |
| "valid_targets_mean": 6764.2, | |
| "valid_targets_min": 5046 | |
| }, | |
| { | |
| "epoch": 4.29994209612044, | |
| "grad_norm": 0.4901600888939512, | |
| "learning_rate": 1.556405221100536e-05, | |
| "loss": 0.176, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09228479862213135, | |
| "step": 3715, | |
| "valid_targets_mean": 6055.8, | |
| "valid_targets_min": 5182 | |
| }, | |
| { | |
| "epoch": 4.305732484076433, | |
| "grad_norm": 0.4023050347526533, | |
| "learning_rate": 1.5507790322337087e-05, | |
| "loss": 0.1707, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08523489534854889, | |
| "step": 3720, | |
| "valid_targets_mean": 7101.6, | |
| "valid_targets_min": 5366 | |
| }, | |
| { | |
| "epoch": 4.311522872032426, | |
| "grad_norm": 0.429715793834325, | |
| "learning_rate": 1.5451565846691997e-05, | |
| "loss": 0.1666, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0880584716796875, | |
| "step": 3725, | |
| "valid_targets_mean": 7111.9, | |
| "valid_targets_min": 5192 | |
| }, | |
| { | |
| "epoch": 4.31731325998842, | |
| "grad_norm": 0.47100851970652413, | |
| "learning_rate": 1.5395379252331323e-05, | |
| "loss": 0.1712, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08530952036380768, | |
| "step": 3730, | |
| "valid_targets_mean": 5509.9, | |
| "valid_targets_min": 4790 | |
| }, | |
| { | |
| "epoch": 4.3231036479444125, | |
| "grad_norm": 0.7478325913135395, | |
| "learning_rate": 1.533923100720081e-05, | |
| "loss": 0.1871, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14090920984745026, | |
| "step": 3735, | |
| "valid_targets_mean": 3874.2, | |
| "valid_targets_min": 1457 | |
| }, | |
| { | |
| "epoch": 4.328894035900405, | |
| "grad_norm": 0.5767617826201782, | |
| "learning_rate": 1.528312157892678e-05, | |
| "loss": 0.233, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09905903041362762, | |
| "step": 3740, | |
| "valid_targets_mean": 3920.0, | |
| "valid_targets_min": 533 | |
| }, | |
| { | |
| "epoch": 4.334684423856398, | |
| "grad_norm": 0.5861299199156408, | |
| "learning_rate": 1.5227051434812317e-05, | |
| "loss": 0.2307, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10419026762247086, | |
| "step": 3745, | |
| "valid_targets_mean": 4077.0, | |
| "valid_targets_min": 1752 | |
| }, | |
| { | |
| "epoch": 4.340474811812391, | |
| "grad_norm": 0.60786331148038, | |
| "learning_rate": 1.5171021041833294e-05, | |
| "loss": 0.225, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14355988800525665, | |
| "step": 3750, | |
| "valid_targets_mean": 4813.5, | |
| "valid_targets_min": 1126 | |
| }, | |
| { | |
| "epoch": 4.346265199768385, | |
| "grad_norm": 0.5655458438425296, | |
| "learning_rate": 1.5115030866634542e-05, | |
| "loss": 0.2173, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10995228588581085, | |
| "step": 3755, | |
| "valid_targets_mean": 4727.9, | |
| "valid_targets_min": 3308 | |
| }, | |
| { | |
| "epoch": 4.352055587724378, | |
| "grad_norm": 0.665312876122124, | |
| "learning_rate": 1.5059081375525937e-05, | |
| "loss": 0.2349, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14005246758460999, | |
| "step": 3760, | |
| "valid_targets_mean": 5534.2, | |
| "valid_targets_min": 1432 | |
| }, | |
| { | |
| "epoch": 4.3578459756803705, | |
| "grad_norm": 0.5884469602525774, | |
| "learning_rate": 1.500317303447851e-05, | |
| "loss": 0.2269, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11218597739934921, | |
| "step": 3765, | |
| "valid_targets_mean": 4703.4, | |
| "valid_targets_min": 1541 | |
| }, | |
| { | |
| "epoch": 4.363636363636363, | |
| "grad_norm": 0.530958275217967, | |
| "learning_rate": 1.4947306309120598e-05, | |
| "loss": 0.2248, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13557025790214539, | |
| "step": 3770, | |
| "valid_targets_mean": 6251.0, | |
| "valid_targets_min": 1855 | |
| }, | |
| { | |
| "epoch": 4.369426751592357, | |
| "grad_norm": 0.5772578330417059, | |
| "learning_rate": 1.489148166473391e-05, | |
| "loss": 0.2221, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.083745077252388, | |
| "step": 3775, | |
| "valid_targets_mean": 3304.2, | |
| "valid_targets_min": 1024 | |
| }, | |
| { | |
| "epoch": 4.37521713954835, | |
| "grad_norm": 0.5667875048921107, | |
| "learning_rate": 1.483569956624972e-05, | |
| "loss": 0.2342, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0941530242562294, | |
| "step": 3780, | |
| "valid_targets_mean": 3925.4, | |
| "valid_targets_min": 2309 | |
| }, | |
| { | |
| "epoch": 4.381007527504343, | |
| "grad_norm": 0.6169957824519388, | |
| "learning_rate": 1.4779960478244951e-05, | |
| "loss": 0.2284, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10853932797908783, | |
| "step": 3785, | |
| "valid_targets_mean": 3988.9, | |
| "valid_targets_min": 1401 | |
| }, | |
| { | |
| "epoch": 4.386797915460336, | |
| "grad_norm": 0.44807236365204745, | |
| "learning_rate": 1.472426486493831e-05, | |
| "loss": 0.2197, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12988603115081787, | |
| "step": 3790, | |
| "valid_targets_mean": 8196.8, | |
| "valid_targets_min": 3317 | |
| }, | |
| { | |
| "epoch": 4.392588303416328, | |
| "grad_norm": 3.0327013394529097, | |
| "learning_rate": 1.4668613190186436e-05, | |
| "loss": 0.2278, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11648248136043549, | |
| "step": 3795, | |
| "valid_targets_mean": 4238.2, | |
| "valid_targets_min": 1650 | |
| }, | |
| { | |
| "epoch": 4.398378691372322, | |
| "grad_norm": 0.5537086119802135, | |
| "learning_rate": 1.4613005917480031e-05, | |
| "loss": 0.2224, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10524934530258179, | |
| "step": 3800, | |
| "valid_targets_mean": 4466.8, | |
| "valid_targets_min": 1412 | |
| }, | |
| { | |
| "epoch": 4.404169079328315, | |
| "grad_norm": 0.6147355430304815, | |
| "learning_rate": 1.4557443509939994e-05, | |
| "loss": 0.2198, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10133112221956253, | |
| "step": 3805, | |
| "valid_targets_mean": 3129.1, | |
| "valid_targets_min": 1354 | |
| }, | |
| { | |
| "epoch": 4.409959467284308, | |
| "grad_norm": 0.5531691642814615, | |
| "learning_rate": 1.4501926430313563e-05, | |
| "loss": 0.2219, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10187403112649918, | |
| "step": 3810, | |
| "valid_targets_mean": 4856.8, | |
| "valid_targets_min": 2052 | |
| }, | |
| { | |
| "epoch": 4.415749855240301, | |
| "grad_norm": 0.6814516561894316, | |
| "learning_rate": 1.4446455140970463e-05, | |
| "loss": 0.2136, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09588038921356201, | |
| "step": 3815, | |
| "valid_targets_mean": 3705.2, | |
| "valid_targets_min": 1210 | |
| }, | |
| { | |
| "epoch": 4.4215402431962945, | |
| "grad_norm": 0.6617520741751552, | |
| "learning_rate": 1.439103010389908e-05, | |
| "loss": 0.2226, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15332624316215515, | |
| "step": 3820, | |
| "valid_targets_mean": 4497.4, | |
| "valid_targets_min": 1293 | |
| }, | |
| { | |
| "epoch": 4.427330631152287, | |
| "grad_norm": 0.6358424187335716, | |
| "learning_rate": 1.4335651780702571e-05, | |
| "loss": 0.2302, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12418112903833389, | |
| "step": 3825, | |
| "valid_targets_mean": 4081.0, | |
| "valid_targets_min": 2979 | |
| }, | |
| { | |
| "epoch": 4.43312101910828, | |
| "grad_norm": 0.6854443786877813, | |
| "learning_rate": 1.4280320632595045e-05, | |
| "loss": 0.2273, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1276114284992218, | |
| "step": 3830, | |
| "valid_targets_mean": 4278.9, | |
| "valid_targets_min": 671 | |
| }, | |
| { | |
| "epoch": 4.438911407064273, | |
| "grad_norm": 0.7096522105030221, | |
| "learning_rate": 1.422503712039772e-05, | |
| "loss": 0.2246, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08994592726230621, | |
| "step": 3835, | |
| "valid_targets_mean": 2743.2, | |
| "valid_targets_min": 1291 | |
| }, | |
| { | |
| "epoch": 4.444701795020267, | |
| "grad_norm": 0.6418924276012431, | |
| "learning_rate": 1.4169801704535089e-05, | |
| "loss": 0.2075, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09683261066675186, | |
| "step": 3840, | |
| "valid_targets_mean": 2747.0, | |
| "valid_targets_min": 1285 | |
| }, | |
| { | |
| "epoch": 4.45049218297626, | |
| "grad_norm": 0.6820132674699142, | |
| "learning_rate": 1.4114614845031056e-05, | |
| "loss": 0.2181, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11021986603736877, | |
| "step": 3845, | |
| "valid_targets_mean": 3813.1, | |
| "valid_targets_min": 1209 | |
| }, | |
| { | |
| "epoch": 4.456282570932252, | |
| "grad_norm": 0.6767122819071981, | |
| "learning_rate": 1.4059477001505142e-05, | |
| "loss": 0.2006, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11383691430091858, | |
| "step": 3850, | |
| "valid_targets_mean": 4358.9, | |
| "valid_targets_min": 3280 | |
| }, | |
| { | |
| "epoch": 4.462072958888245, | |
| "grad_norm": 0.648493463933508, | |
| "learning_rate": 1.4004388633168644e-05, | |
| "loss": 0.2256, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08743107318878174, | |
| "step": 3855, | |
| "valid_targets_mean": 3406.6, | |
| "valid_targets_min": 1957 | |
| }, | |
| { | |
| "epoch": 4.467863346844238, | |
| "grad_norm": 0.6544777379199935, | |
| "learning_rate": 1.3949350198820805e-05, | |
| "loss": 0.224, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15421587228775024, | |
| "step": 3860, | |
| "valid_targets_mean": 4873.1, | |
| "valid_targets_min": 2833 | |
| }, | |
| { | |
| "epoch": 4.473653734800232, | |
| "grad_norm": 0.6873146316651686, | |
| "learning_rate": 1.3894362156845004e-05, | |
| "loss": 0.2084, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08594775944948196, | |
| "step": 3865, | |
| "valid_targets_mean": 2926.9, | |
| "valid_targets_min": 1846 | |
| }, | |
| { | |
| "epoch": 4.479444122756225, | |
| "grad_norm": 0.6697088686410119, | |
| "learning_rate": 1.3839424965204913e-05, | |
| "loss": 0.2159, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09900952130556107, | |
| "step": 3870, | |
| "valid_targets_mean": 3012.8, | |
| "valid_targets_min": 1009 | |
| }, | |
| { | |
| "epoch": 4.485234510712218, | |
| "grad_norm": 0.6780954186588826, | |
| "learning_rate": 1.3784539081440729e-05, | |
| "loss": 0.2084, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10792041569948196, | |
| "step": 3875, | |
| "valid_targets_mean": 3220.8, | |
| "valid_targets_min": 1264 | |
| }, | |
| { | |
| "epoch": 4.49102489866821, | |
| "grad_norm": 0.5955939245300237, | |
| "learning_rate": 1.3729704962665294e-05, | |
| "loss": 0.2126, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09530109167098999, | |
| "step": 3880, | |
| "valid_targets_mean": 3489.0, | |
| "valid_targets_min": 1606 | |
| }, | |
| { | |
| "epoch": 4.496815286624204, | |
| "grad_norm": 0.7449723707975259, | |
| "learning_rate": 1.3674923065560375e-05, | |
| "loss": 0.2156, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10313093662261963, | |
| "step": 3885, | |
| "valid_targets_mean": 3216.2, | |
| "valid_targets_min": 1190 | |
| }, | |
| { | |
| "epoch": 4.502605674580197, | |
| "grad_norm": 0.681314779850815, | |
| "learning_rate": 1.3620193846372778e-05, | |
| "loss": 0.2076, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1107458770275116, | |
| "step": 3890, | |
| "valid_targets_mean": 3226.9, | |
| "valid_targets_min": 571 | |
| }, | |
| { | |
| "epoch": 4.50839606253619, | |
| "grad_norm": 0.6618093464158726, | |
| "learning_rate": 1.3565517760910605e-05, | |
| "loss": 0.2106, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11724066734313965, | |
| "step": 3895, | |
| "valid_targets_mean": 4088.4, | |
| "valid_targets_min": 2410 | |
| }, | |
| { | |
| "epoch": 4.514186450492183, | |
| "grad_norm": 0.6208808503309192, | |
| "learning_rate": 1.3510895264539446e-05, | |
| "loss": 0.2109, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09507305175065994, | |
| "step": 3900, | |
| "valid_targets_mean": 3610.4, | |
| "valid_targets_min": 1110 | |
| }, | |
| { | |
| "epoch": 4.519976838448176, | |
| "grad_norm": 0.7699986741181999, | |
| "learning_rate": 1.3456326812178546e-05, | |
| "loss": 0.2057, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10935103893280029, | |
| "step": 3905, | |
| "valid_targets_mean": 3476.2, | |
| "valid_targets_min": 513 | |
| }, | |
| { | |
| "epoch": 4.525767226404169, | |
| "grad_norm": 0.7441197830906147, | |
| "learning_rate": 1.3401812858297087e-05, | |
| "loss": 0.2085, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1089061051607132, | |
| "step": 3910, | |
| "valid_targets_mean": 4105.5, | |
| "valid_targets_min": 2558 | |
| }, | |
| { | |
| "epoch": 4.531557614360162, | |
| "grad_norm": 0.6779694865278227, | |
| "learning_rate": 1.334735385691033e-05, | |
| "loss": 0.2232, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11115449666976929, | |
| "step": 3915, | |
| "valid_targets_mean": 3982.9, | |
| "valid_targets_min": 3108 | |
| }, | |
| { | |
| "epoch": 4.537348002316155, | |
| "grad_norm": 0.6526369552324457, | |
| "learning_rate": 1.329295026157589e-05, | |
| "loss": 0.2105, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09470139443874359, | |
| "step": 3920, | |
| "valid_targets_mean": 3392.6, | |
| "valid_targets_min": 1860 | |
| }, | |
| { | |
| "epoch": 4.543138390272148, | |
| "grad_norm": 0.5847112270187543, | |
| "learning_rate": 1.3238602525389937e-05, | |
| "loss": 0.2125, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12164326012134552, | |
| "step": 3925, | |
| "valid_targets_mean": 5552.2, | |
| "valid_targets_min": 2495 | |
| }, | |
| { | |
| "epoch": 4.548928778228142, | |
| "grad_norm": 0.5814037340512235, | |
| "learning_rate": 1.3184311100983407e-05, | |
| "loss": 0.2085, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09199385344982147, | |
| "step": 3930, | |
| "valid_targets_mean": 3295.2, | |
| "valid_targets_min": 1292 | |
| }, | |
| { | |
| "epoch": 4.554719166184134, | |
| "grad_norm": 0.6984013547866225, | |
| "learning_rate": 1.313007644051826e-05, | |
| "loss": 0.2212, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11497661471366882, | |
| "step": 3935, | |
| "valid_targets_mean": 3739.6, | |
| "valid_targets_min": 1655 | |
| }, | |
| { | |
| "epoch": 4.560509554140127, | |
| "grad_norm": 0.6552364897317641, | |
| "learning_rate": 1.3075898995683707e-05, | |
| "loss": 0.206, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1044774204492569, | |
| "step": 3940, | |
| "valid_targets_mean": 3523.8, | |
| "valid_targets_min": 1757 | |
| }, | |
| { | |
| "epoch": 4.56629994209612, | |
| "grad_norm": 0.6855739539292031, | |
| "learning_rate": 1.3021779217692432e-05, | |
| "loss": 0.2108, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1120901107788086, | |
| "step": 3945, | |
| "valid_targets_mean": 4563.8, | |
| "valid_targets_min": 2313 | |
| }, | |
| { | |
| "epoch": 4.572090330052114, | |
| "grad_norm": 0.6724289348839543, | |
| "learning_rate": 1.2967717557276841e-05, | |
| "loss": 0.2125, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09848484396934509, | |
| "step": 3950, | |
| "valid_targets_mean": 3368.1, | |
| "valid_targets_min": 2192 | |
| }, | |
| { | |
| "epoch": 4.577880718008107, | |
| "grad_norm": 0.6404908890629426, | |
| "learning_rate": 1.2913714464685322e-05, | |
| "loss": 0.2064, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0937051847577095, | |
| "step": 3955, | |
| "valid_targets_mean": 3490.8, | |
| "valid_targets_min": 1833 | |
| }, | |
| { | |
| "epoch": 4.5836711059640995, | |
| "grad_norm": 0.7226581927394442, | |
| "learning_rate": 1.2859770389678485e-05, | |
| "loss": 0.2069, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09981274604797363, | |
| "step": 3960, | |
| "valid_targets_mean": 2947.1, | |
| "valid_targets_min": 1200 | |
| }, | |
| { | |
| "epoch": 4.589461493920092, | |
| "grad_norm": 0.6019247172317036, | |
| "learning_rate": 1.2805885781525418e-05, | |
| "loss": 0.2138, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10194752365350723, | |
| "step": 3965, | |
| "valid_targets_mean": 5060.6, | |
| "valid_targets_min": 3501 | |
| }, | |
| { | |
| "epoch": 4.595251881876086, | |
| "grad_norm": 0.6650891905748428, | |
| "learning_rate": 1.2752061088999935e-05, | |
| "loss": 0.2056, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1218477189540863, | |
| "step": 3970, | |
| "valid_targets_mean": 3855.6, | |
| "valid_targets_min": 2498 | |
| }, | |
| { | |
| "epoch": 4.601042269832079, | |
| "grad_norm": 0.6358089854292314, | |
| "learning_rate": 1.2698296760376851e-05, | |
| "loss": 0.2183, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09213036298751831, | |
| "step": 3975, | |
| "valid_targets_mean": 3354.5, | |
| "valid_targets_min": 988 | |
| }, | |
| { | |
| "epoch": 4.606832657788072, | |
| "grad_norm": 0.7708131148847596, | |
| "learning_rate": 1.264459324342826e-05, | |
| "loss": 0.2063, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11673158407211304, | |
| "step": 3980, | |
| "valid_targets_mean": 3658.9, | |
| "valid_targets_min": 1361 | |
| }, | |
| { | |
| "epoch": 4.612623045744065, | |
| "grad_norm": 0.6343915324436624, | |
| "learning_rate": 1.2590950985419759e-05, | |
| "loss": 0.2178, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11010897159576416, | |
| "step": 3985, | |
| "valid_targets_mean": 3907.2, | |
| "valid_targets_min": 1685 | |
| }, | |
| { | |
| "epoch": 4.6184134337000575, | |
| "grad_norm": 0.7641387693225504, | |
| "learning_rate": 1.2537370433106783e-05, | |
| "loss": 0.2093, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09839899837970734, | |
| "step": 3990, | |
| "valid_targets_mean": 2536.4, | |
| "valid_targets_min": 474 | |
| }, | |
| { | |
| "epoch": 4.624203821656051, | |
| "grad_norm": 0.6700502378999571, | |
| "learning_rate": 1.2483852032730843e-05, | |
| "loss": 0.209, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1001468151807785, | |
| "step": 3995, | |
| "valid_targets_mean": 3377.9, | |
| "valid_targets_min": 2200 | |
| }, | |
| { | |
| "epoch": 4.629994209612044, | |
| "grad_norm": 0.6452149033447232, | |
| "learning_rate": 1.2430396230015825e-05, | |
| "loss": 0.2015, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10308694839477539, | |
| "step": 4000, | |
| "valid_targets_mean": 4266.8, | |
| "valid_targets_min": 2241 | |
| }, | |
| { | |
| "epoch": 4.635784597568037, | |
| "grad_norm": 0.6567104338943586, | |
| "learning_rate": 1.2377003470164279e-05, | |
| "loss": 0.2043, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13291510939598083, | |
| "step": 4005, | |
| "valid_targets_mean": 4576.9, | |
| "valid_targets_min": 1478 | |
| }, | |
| { | |
| "epoch": 4.64157498552403, | |
| "grad_norm": 1.1276081511241993, | |
| "learning_rate": 1.2323674197853697e-05, | |
| "loss": 0.1898, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09665089100599289, | |
| "step": 4010, | |
| "valid_targets_mean": 4326.5, | |
| "valid_targets_min": 1194 | |
| }, | |
| { | |
| "epoch": 4.6473653734800235, | |
| "grad_norm": 0.658002633108222, | |
| "learning_rate": 1.2270408857232834e-05, | |
| "loss": 0.2232, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11016303300857544, | |
| "step": 4015, | |
| "valid_targets_mean": 3412.9, | |
| "valid_targets_min": 1220 | |
| }, | |
| { | |
| "epoch": 4.653155761436016, | |
| "grad_norm": 0.618944368962821, | |
| "learning_rate": 1.221720789191798e-05, | |
| "loss": 0.2008, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09453439712524414, | |
| "step": 4020, | |
| "valid_targets_mean": 4365.8, | |
| "valid_targets_min": 1988 | |
| }, | |
| { | |
| "epoch": 4.658946149392009, | |
| "grad_norm": 0.6190869503731473, | |
| "learning_rate": 1.2164071744989294e-05, | |
| "loss": 0.2134, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09763002395629883, | |
| "step": 4025, | |
| "valid_targets_mean": 3706.8, | |
| "valid_targets_min": 1256 | |
| }, | |
| { | |
| "epoch": 4.664736537348002, | |
| "grad_norm": 0.7315105123119252, | |
| "learning_rate": 1.2111000858987082e-05, | |
| "loss": 0.2088, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1276649832725525, | |
| "step": 4030, | |
| "valid_targets_mean": 3325.8, | |
| "valid_targets_min": 1608 | |
| }, | |
| { | |
| "epoch": 4.670526925303996, | |
| "grad_norm": 0.6708486543908649, | |
| "learning_rate": 1.2057995675908153e-05, | |
| "loss": 0.2192, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11391855031251907, | |
| "step": 4035, | |
| "valid_targets_mean": 4343.9, | |
| "valid_targets_min": 2800 | |
| }, | |
| { | |
| "epoch": 4.676317313259989, | |
| "grad_norm": 0.6927673324719886, | |
| "learning_rate": 1.2005056637202098e-05, | |
| "loss": 0.2085, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10946047306060791, | |
| "step": 4040, | |
| "valid_targets_mean": 3786.2, | |
| "valid_targets_min": 1771 | |
| }, | |
| { | |
| "epoch": 4.6821077012159815, | |
| "grad_norm": 0.6821074198825163, | |
| "learning_rate": 1.1952184183767638e-05, | |
| "loss": 0.1997, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11058327555656433, | |
| "step": 4045, | |
| "valid_targets_mean": 3600.8, | |
| "valid_targets_min": 2528 | |
| }, | |
| { | |
| "epoch": 4.687898089171974, | |
| "grad_norm": 0.7512276074888579, | |
| "learning_rate": 1.1899378755948931e-05, | |
| "loss": 0.1973, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0803704783320427, | |
| "step": 4050, | |
| "valid_targets_mean": 2860.0, | |
| "valid_targets_min": 854 | |
| }, | |
| { | |
| "epoch": 4.693688477127967, | |
| "grad_norm": 0.7374796362496391, | |
| "learning_rate": 1.1846640793531941e-05, | |
| "loss": 0.2026, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11275653541088104, | |
| "step": 4055, | |
| "valid_targets_mean": 3152.2, | |
| "valid_targets_min": 2343 | |
| }, | |
| { | |
| "epoch": 4.699478865083961, | |
| "grad_norm": 0.7851951138416005, | |
| "learning_rate": 1.1793970735740716e-05, | |
| "loss": 0.1902, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07852757722139359, | |
| "step": 4060, | |
| "valid_targets_mean": 2654.6, | |
| "valid_targets_min": 1026 | |
| }, | |
| { | |
| "epoch": 4.705269253039954, | |
| "grad_norm": 0.6767815923385693, | |
| "learning_rate": 1.1741369021233799e-05, | |
| "loss": 0.1959, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10796316713094711, | |
| "step": 4065, | |
| "valid_targets_mean": 4003.4, | |
| "valid_targets_min": 1524 | |
| }, | |
| { | |
| "epoch": 4.711059640995947, | |
| "grad_norm": 0.697623902564916, | |
| "learning_rate": 1.1688836088100524e-05, | |
| "loss": 0.1864, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10594123601913452, | |
| "step": 4070, | |
| "valid_targets_mean": 3186.2, | |
| "valid_targets_min": 1274 | |
| }, | |
| { | |
| "epoch": 4.7168500289519395, | |
| "grad_norm": 0.7168919011091767, | |
| "learning_rate": 1.1636372373857388e-05, | |
| "loss": 0.1983, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12310782074928284, | |
| "step": 4075, | |
| "valid_targets_mean": 4443.4, | |
| "valid_targets_min": 914 | |
| }, | |
| { | |
| "epoch": 4.722640416907932, | |
| "grad_norm": 0.6519683034317412, | |
| "learning_rate": 1.1583978315444405e-05, | |
| "loss": 0.2245, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1027081310749054, | |
| "step": 4080, | |
| "valid_targets_mean": 4349.2, | |
| "valid_targets_min": 2093 | |
| }, | |
| { | |
| "epoch": 4.728430804863926, | |
| "grad_norm": 0.6867968732060586, | |
| "learning_rate": 1.1531654349221468e-05, | |
| "loss": 0.2002, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11880075186491013, | |
| "step": 4085, | |
| "valid_targets_mean": 3341.9, | |
| "valid_targets_min": 1621 | |
| }, | |
| { | |
| "epoch": 4.734221192819919, | |
| "grad_norm": 0.6400500919623512, | |
| "learning_rate": 1.147940091096472e-05, | |
| "loss": 0.2056, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0961160808801651, | |
| "step": 4090, | |
| "valid_targets_mean": 4046.6, | |
| "valid_targets_min": 2253 | |
| }, | |
| { | |
| "epoch": 4.740011580775912, | |
| "grad_norm": 0.6612830128083482, | |
| "learning_rate": 1.1427218435862884e-05, | |
| "loss": 0.1978, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08661623299121857, | |
| "step": 4095, | |
| "valid_targets_mean": 4462.8, | |
| "valid_targets_min": 1789 | |
| }, | |
| { | |
| "epoch": 4.7458019687319055, | |
| "grad_norm": 0.6781757851379322, | |
| "learning_rate": 1.1375107358513717e-05, | |
| "loss": 0.2128, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11982538551092148, | |
| "step": 4100, | |
| "valid_targets_mean": 4629.9, | |
| "valid_targets_min": 1586 | |
| }, | |
| { | |
| "epoch": 4.751592356687898, | |
| "grad_norm": 0.7051102633354648, | |
| "learning_rate": 1.1323068112920304e-05, | |
| "loss": 0.2032, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10887384414672852, | |
| "step": 4105, | |
| "valid_targets_mean": 3739.8, | |
| "valid_targets_min": 2391 | |
| }, | |
| { | |
| "epoch": 4.757382744643891, | |
| "grad_norm": 0.590000679776256, | |
| "learning_rate": 1.1271101132487514e-05, | |
| "loss": 0.1945, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08773195743560791, | |
| "step": 4110, | |
| "valid_targets_mean": 3735.9, | |
| "valid_targets_min": 2513 | |
| }, | |
| { | |
| "epoch": 4.763173132599884, | |
| "grad_norm": 0.590068763568892, | |
| "learning_rate": 1.1219206850018351e-05, | |
| "loss": 0.2067, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.081422820687294, | |
| "step": 4115, | |
| "valid_targets_mean": 3797.2, | |
| "valid_targets_min": 2156 | |
| }, | |
| { | |
| "epoch": 4.768963520555877, | |
| "grad_norm": 0.6529370026097409, | |
| "learning_rate": 1.1167385697710361e-05, | |
| "loss": 0.196, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0942734032869339, | |
| "step": 4120, | |
| "valid_targets_mean": 3446.1, | |
| "valid_targets_min": 1423 | |
| }, | |
| { | |
| "epoch": 4.774753908511871, | |
| "grad_norm": 0.7988272664256585, | |
| "learning_rate": 1.1115638107152034e-05, | |
| "loss": 0.1996, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10448580980300903, | |
| "step": 4125, | |
| "valid_targets_mean": 3166.9, | |
| "valid_targets_min": 2126 | |
| }, | |
| { | |
| "epoch": 4.7805442964678635, | |
| "grad_norm": 2.4716700067552764, | |
| "learning_rate": 1.1063964509319188e-05, | |
| "loss": 0.2172, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10709042847156525, | |
| "step": 4130, | |
| "valid_targets_mean": 3749.2, | |
| "valid_targets_min": 1834 | |
| }, | |
| { | |
| "epoch": 4.786334684423856, | |
| "grad_norm": 0.7725124051827306, | |
| "learning_rate": 1.1012365334571432e-05, | |
| "loss": 0.2061, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1110750138759613, | |
| "step": 4135, | |
| "valid_targets_mean": 4225.4, | |
| "valid_targets_min": 1571 | |
| }, | |
| { | |
| "epoch": 4.792125072379849, | |
| "grad_norm": 0.6821783963541409, | |
| "learning_rate": 1.0960841012648506e-05, | |
| "loss": 0.217, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10442662984132767, | |
| "step": 4140, | |
| "valid_targets_mean": 3508.6, | |
| "valid_targets_min": 2150 | |
| }, | |
| { | |
| "epoch": 4.797915460335842, | |
| "grad_norm": 0.6237896593022638, | |
| "learning_rate": 1.0909391972666783e-05, | |
| "loss": 0.2077, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10406245291233063, | |
| "step": 4145, | |
| "valid_targets_mean": 4714.1, | |
| "valid_targets_min": 2761 | |
| }, | |
| { | |
| "epoch": 4.803705848291836, | |
| "grad_norm": 0.6539511811303761, | |
| "learning_rate": 1.0858018643115636e-05, | |
| "loss": 0.2207, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12043410539627075, | |
| "step": 4150, | |
| "valid_targets_mean": 4161.5, | |
| "valid_targets_min": 2492 | |
| }, | |
| { | |
| "epoch": 4.809496236247829, | |
| "grad_norm": 0.6695364647428075, | |
| "learning_rate": 1.0806721451853892e-05, | |
| "loss": 0.2134, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09142296016216278, | |
| "step": 4155, | |
| "valid_targets_mean": 3178.5, | |
| "valid_targets_min": 1675 | |
| }, | |
| { | |
| "epoch": 4.8152866242038215, | |
| "grad_norm": 0.6371529714283551, | |
| "learning_rate": 1.075550082610628e-05, | |
| "loss": 0.2148, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08996136486530304, | |
| "step": 4160, | |
| "valid_targets_mean": 3115.8, | |
| "valid_targets_min": 1868 | |
| }, | |
| { | |
| "epoch": 4.821077012159814, | |
| "grad_norm": 0.7399760446823987, | |
| "learning_rate": 1.0704357192459824e-05, | |
| "loss": 0.2084, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12434670329093933, | |
| "step": 4165, | |
| "valid_targets_mean": 3591.2, | |
| "valid_targets_min": 1969 | |
| }, | |
| { | |
| "epoch": 4.826867400115808, | |
| "grad_norm": 0.6338640690724339, | |
| "learning_rate": 1.065329097686036e-05, | |
| "loss": 0.2104, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09719884395599365, | |
| "step": 4170, | |
| "valid_targets_mean": 3810.2, | |
| "valid_targets_min": 1878 | |
| }, | |
| { | |
| "epoch": 4.832657788071801, | |
| "grad_norm": 0.7371351517240505, | |
| "learning_rate": 1.0602302604608949e-05, | |
| "loss": 0.2122, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10729122161865234, | |
| "step": 4175, | |
| "valid_targets_mean": 3476.5, | |
| "valid_targets_min": 1207 | |
| }, | |
| { | |
| "epoch": 4.838448176027794, | |
| "grad_norm": 0.6493557233873885, | |
| "learning_rate": 1.0551392500358317e-05, | |
| "loss": 0.2087, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10543425381183624, | |
| "step": 4180, | |
| "valid_targets_mean": 4223.1, | |
| "valid_targets_min": 2305 | |
| }, | |
| { | |
| "epoch": 4.844238563983787, | |
| "grad_norm": 0.6871112372096738, | |
| "learning_rate": 1.0500561088109364e-05, | |
| "loss": 0.203, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08310510963201523, | |
| "step": 4185, | |
| "valid_targets_mean": 2648.1, | |
| "valid_targets_min": 953 | |
| }, | |
| { | |
| "epoch": 4.85002895193978, | |
| "grad_norm": 0.6055841246914735, | |
| "learning_rate": 1.04498087912076e-05, | |
| "loss": 0.1958, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09668581187725067, | |
| "step": 4190, | |
| "valid_targets_mean": 5047.8, | |
| "valid_targets_min": 1659 | |
| }, | |
| { | |
| "epoch": 4.855819339895773, | |
| "grad_norm": 0.6417252887209771, | |
| "learning_rate": 1.039913603233964e-05, | |
| "loss": 0.2125, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08402392268180847, | |
| "step": 4195, | |
| "valid_targets_mean": 3317.6, | |
| "valid_targets_min": 1390 | |
| }, | |
| { | |
| "epoch": 4.861609727851766, | |
| "grad_norm": 0.685651764111974, | |
| "learning_rate": 1.0348543233529647e-05, | |
| "loss": 0.2004, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10304082930088043, | |
| "step": 4200, | |
| "valid_targets_mean": 3857.2, | |
| "valid_targets_min": 2032 | |
| }, | |
| { | |
| "epoch": 4.867400115807759, | |
| "grad_norm": 0.6221040398072666, | |
| "learning_rate": 1.0298030816135866e-05, | |
| "loss": 0.2136, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09167119860649109, | |
| "step": 4205, | |
| "valid_targets_mean": 3392.9, | |
| "valid_targets_min": 1464 | |
| }, | |
| { | |
| "epoch": 4.873190503763752, | |
| "grad_norm": 0.7112812688595366, | |
| "learning_rate": 1.024759920084709e-05, | |
| "loss": 0.2075, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10534010827541351, | |
| "step": 4210, | |
| "valid_targets_mean": 2917.6, | |
| "valid_targets_min": 1826 | |
| }, | |
| { | |
| "epoch": 4.8789808917197455, | |
| "grad_norm": 0.6305949802947574, | |
| "learning_rate": 1.0197248807679152e-05, | |
| "loss": 0.2111, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10810261964797974, | |
| "step": 4215, | |
| "valid_targets_mean": 4448.1, | |
| "valid_targets_min": 2900 | |
| }, | |
| { | |
| "epoch": 4.884771279675738, | |
| "grad_norm": 0.7233919857839544, | |
| "learning_rate": 1.0146980055971422e-05, | |
| "loss": 0.2029, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11249633878469467, | |
| "step": 4220, | |
| "valid_targets_mean": 3419.1, | |
| "valid_targets_min": 2194 | |
| }, | |
| { | |
| "epoch": 4.890561667631731, | |
| "grad_norm": 0.6491504770527533, | |
| "learning_rate": 1.0096793364383346e-05, | |
| "loss": 0.2137, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10216433554887772, | |
| "step": 4225, | |
| "valid_targets_mean": 4196.8, | |
| "valid_targets_min": 1931 | |
| }, | |
| { | |
| "epoch": 4.896352055587724, | |
| "grad_norm": 0.5979581538813098, | |
| "learning_rate": 1.0046689150890935e-05, | |
| "loss": 0.2066, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09729495644569397, | |
| "step": 4230, | |
| "valid_targets_mean": 5424.6, | |
| "valid_targets_min": 1233 | |
| }, | |
| { | |
| "epoch": 4.902142443543718, | |
| "grad_norm": 0.5635032941313716, | |
| "learning_rate": 9.996667832783266e-06, | |
| "loss": 0.1997, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10651378333568573, | |
| "step": 4235, | |
| "valid_targets_mean": 5272.1, | |
| "valid_targets_min": 3440 | |
| }, | |
| { | |
| "epoch": 4.907932831499711, | |
| "grad_norm": 0.6161802350596663, | |
| "learning_rate": 9.946729826659056e-06, | |
| "loss": 0.2001, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09289420396089554, | |
| "step": 4240, | |
| "valid_targets_mean": 3633.9, | |
| "valid_targets_min": 1873 | |
| }, | |
| { | |
| "epoch": 4.9137232194557035, | |
| "grad_norm": 0.6419586212420326, | |
| "learning_rate": 9.896875548423145e-06, | |
| "loss": 0.1996, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09963817894458771, | |
| "step": 4245, | |
| "valid_targets_mean": 3971.1, | |
| "valid_targets_min": 1614 | |
| }, | |
| { | |
| "epoch": 4.919513607411696, | |
| "grad_norm": 0.6527250029035306, | |
| "learning_rate": 9.847105413283067e-06, | |
| "loss": 0.2024, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11336039006710052, | |
| "step": 4250, | |
| "valid_targets_mean": 4600.5, | |
| "valid_targets_min": 974 | |
| }, | |
| { | |
| "epoch": 4.92530399536769, | |
| "grad_norm": 0.6875372467737946, | |
| "learning_rate": 9.797419835745572e-06, | |
| "loss": 0.1977, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09850144386291504, | |
| "step": 4255, | |
| "valid_targets_mean": 3246.1, | |
| "valid_targets_min": 2054 | |
| }, | |
| { | |
| "epoch": 4.931094383323683, | |
| "grad_norm": 0.6776141150310618, | |
| "learning_rate": 9.747819229613163e-06, | |
| "loss": 0.1965, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10144221782684326, | |
| "step": 4260, | |
| "valid_targets_mean": 4019.6, | |
| "valid_targets_min": 1333 | |
| }, | |
| { | |
| "epoch": 4.936884771279676, | |
| "grad_norm": 0.7006395388235903, | |
| "learning_rate": 9.698304007980688e-06, | |
| "loss": 0.2013, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08913317322731018, | |
| "step": 4265, | |
| "valid_targets_mean": 3301.9, | |
| "valid_targets_min": 1515 | |
| }, | |
| { | |
| "epoch": 4.942675159235669, | |
| "grad_norm": 0.733734340115638, | |
| "learning_rate": 9.648874583231858e-06, | |
| "loss": 0.2152, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11472181975841522, | |
| "step": 4270, | |
| "valid_targets_mean": 2966.4, | |
| "valid_targets_min": 1349 | |
| }, | |
| { | |
| "epoch": 4.948465547191661, | |
| "grad_norm": 0.6786184673726235, | |
| "learning_rate": 9.59953136703584e-06, | |
| "loss": 0.2131, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10117700695991516, | |
| "step": 4275, | |
| "valid_targets_mean": 3284.2, | |
| "valid_targets_min": 1906 | |
| }, | |
| { | |
| "epoch": 4.954255935147655, | |
| "grad_norm": 0.7093575427420685, | |
| "learning_rate": 9.550274770343826e-06, | |
| "loss": 0.2019, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11812548339366913, | |
| "step": 4280, | |
| "valid_targets_mean": 3551.6, | |
| "valid_targets_min": 2076 | |
| }, | |
| { | |
| "epoch": 4.960046323103648, | |
| "grad_norm": 0.6242435134589384, | |
| "learning_rate": 9.501105203385592e-06, | |
| "loss": 0.194, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09157514572143555, | |
| "step": 4285, | |
| "valid_targets_mean": 3506.8, | |
| "valid_targets_min": 1021 | |
| }, | |
| { | |
| "epoch": 4.965836711059641, | |
| "grad_norm": 0.7343590166234107, | |
| "learning_rate": 9.452023075666104e-06, | |
| "loss": 0.2087, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1187494695186615, | |
| "step": 4290, | |
| "valid_targets_mean": 3709.6, | |
| "valid_targets_min": 2413 | |
| }, | |
| { | |
| "epoch": 4.971627099015634, | |
| "grad_norm": 0.645442375731764, | |
| "learning_rate": 9.403028795962095e-06, | |
| "loss": 0.1983, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09993511438369751, | |
| "step": 4295, | |
| "valid_targets_mean": 3746.6, | |
| "valid_targets_min": 1605 | |
| }, | |
| { | |
| "epoch": 4.9774174869716274, | |
| "grad_norm": 0.7213539538482706, | |
| "learning_rate": 9.354122772318654e-06, | |
| "loss": 0.2103, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08953292667865753, | |
| "step": 4300, | |
| "valid_targets_mean": 2676.8, | |
| "valid_targets_min": 815 | |
| }, | |
| { | |
| "epoch": 4.98320787492762, | |
| "grad_norm": 0.6582835715939438, | |
| "learning_rate": 9.305305412045836e-06, | |
| "loss": 0.2103, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09620843827724457, | |
| "step": 4305, | |
| "valid_targets_mean": 3131.9, | |
| "valid_targets_min": 1878 | |
| }, | |
| { | |
| "epoch": 4.988998262883613, | |
| "grad_norm": 0.6150045532252543, | |
| "learning_rate": 9.256577121715278e-06, | |
| "loss": 0.2043, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10262969136238098, | |
| "step": 4310, | |
| "valid_targets_mean": 4093.5, | |
| "valid_targets_min": 1793 | |
| }, | |
| { | |
| "epoch": 4.994788650839606, | |
| "grad_norm": 0.6397601890340455, | |
| "learning_rate": 9.207938307156808e-06, | |
| "loss": 0.2103, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10116622596979141, | |
| "step": 4315, | |
| "valid_targets_mean": 3702.9, | |
| "valid_targets_min": 1690 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 0.7476582503069477, | |
| "learning_rate": 9.159389373455055e-06, | |
| "loss": 0.2018, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16756142675876617, | |
| "step": 4320, | |
| "valid_targets_mean": 5459.5, | |
| "valid_targets_min": 2092 | |
| }, | |
| { | |
| "epoch": 5.005790387955993, | |
| "grad_norm": 0.5170706009126839, | |
| "learning_rate": 9.11093072494608e-06, | |
| "loss": 0.1807, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08678993582725525, | |
| "step": 4325, | |
| "valid_targets_mean": 7389.0, | |
| "valid_targets_min": 5627 | |
| }, | |
| { | |
| "epoch": 5.011580775911986, | |
| "grad_norm": 0.7348375708355329, | |
| "learning_rate": 9.06256276521402e-06, | |
| "loss": 0.179, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.04887574166059494, | |
| "step": 4330, | |
| "valid_targets_mean": 1294.8, | |
| "valid_targets_min": 422 | |
| }, | |
| { | |
| "epoch": 5.017371163867979, | |
| "grad_norm": 0.40825306898806885, | |
| "learning_rate": 9.014285897087717e-06, | |
| "loss": 0.1718, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09006273746490479, | |
| "step": 4335, | |
| "valid_targets_mean": 8539.9, | |
| "valid_targets_min": 6761 | |
| }, | |
| { | |
| "epoch": 5.023161551823972, | |
| "grad_norm": 0.4169264118454129, | |
| "learning_rate": 8.966100522637349e-06, | |
| "loss": 0.1692, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07914991676807404, | |
| "step": 4340, | |
| "valid_targets_mean": 7579.0, | |
| "valid_targets_min": 6174 | |
| }, | |
| { | |
| "epoch": 5.028951939779965, | |
| "grad_norm": 0.4261254122872559, | |
| "learning_rate": 8.9180070431711e-06, | |
| "loss": 0.1718, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07706992328166962, | |
| "step": 4345, | |
| "valid_targets_mean": 7328.2, | |
| "valid_targets_min": 4839 | |
| }, | |
| { | |
| "epoch": 5.034742327735958, | |
| "grad_norm": 0.49356145166231324, | |
| "learning_rate": 8.870005859231824e-06, | |
| "loss": 0.1809, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09671644866466522, | |
| "step": 4350, | |
| "valid_targets_mean": 6838.2, | |
| "valid_targets_min": 5034 | |
| }, | |
| { | |
| "epoch": 5.040532715691952, | |
| "grad_norm": 0.45057468994612204, | |
| "learning_rate": 8.822097370593699e-06, | |
| "loss": 0.1832, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.092893585562706, | |
| "step": 4355, | |
| "valid_targets_mean": 7248.2, | |
| "valid_targets_min": 5567 | |
| }, | |
| { | |
| "epoch": 5.046323103647945, | |
| "grad_norm": 0.42693244283983495, | |
| "learning_rate": 8.774281976258885e-06, | |
| "loss": 0.1692, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07744898647069931, | |
| "step": 4360, | |
| "valid_targets_mean": 6234.0, | |
| "valid_targets_min": 5603 | |
| }, | |
| { | |
| "epoch": 5.052113491603937, | |
| "grad_norm": 0.7444283539312915, | |
| "learning_rate": 8.726560074454224e-06, | |
| "loss": 0.1641, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.02484450861811638, | |
| "step": 4365, | |
| "valid_targets_mean": 602.0, | |
| "valid_targets_min": 152 | |
| }, | |
| { | |
| "epoch": 5.05790387955993, | |
| "grad_norm": 0.41468949731358445, | |
| "learning_rate": 8.678932062627919e-06, | |
| "loss": 0.1557, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07928498834371567, | |
| "step": 4370, | |
| "valid_targets_mean": 7299.9, | |
| "valid_targets_min": 6158 | |
| }, | |
| { | |
| "epoch": 5.063694267515924, | |
| "grad_norm": 0.4539339242355834, | |
| "learning_rate": 8.63139833744619e-06, | |
| "loss": 0.1716, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08123583346605301, | |
| "step": 4375, | |
| "valid_targets_mean": 6140.0, | |
| "valid_targets_min": 4648 | |
| }, | |
| { | |
| "epoch": 5.069484655471917, | |
| "grad_norm": 0.4534263931745652, | |
| "learning_rate": 8.583959294790024e-06, | |
| "loss": 0.1643, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08489354699850082, | |
| "step": 4380, | |
| "valid_targets_mean": 6603.0, | |
| "valid_targets_min": 4882 | |
| }, | |
| { | |
| "epoch": 5.07527504342791, | |
| "grad_norm": 0.4213647354140185, | |
| "learning_rate": 8.53661532975184e-06, | |
| "loss": 0.1666, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08459046483039856, | |
| "step": 4385, | |
| "valid_targets_mean": 8015.0, | |
| "valid_targets_min": 6344 | |
| }, | |
| { | |
| "epoch": 5.0810654313839025, | |
| "grad_norm": 0.4403748874087747, | |
| "learning_rate": 8.489366836632215e-06, | |
| "loss": 0.1581, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.073912113904953, | |
| "step": 4390, | |
| "valid_targets_mean": 6305.2, | |
| "valid_targets_min": 4622 | |
| }, | |
| { | |
| "epoch": 5.086855819339895, | |
| "grad_norm": 0.4217263365160654, | |
| "learning_rate": 8.442214208936594e-06, | |
| "loss": 0.1671, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08894698321819305, | |
| "step": 4395, | |
| "valid_targets_mean": 7890.8, | |
| "valid_targets_min": 5485 | |
| }, | |
| { | |
| "epoch": 5.092646207295889, | |
| "grad_norm": 0.41608850779826057, | |
| "learning_rate": 8.395157839372017e-06, | |
| "loss": 0.1673, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07635587453842163, | |
| "step": 4400, | |
| "valid_targets_mean": 7016.6, | |
| "valid_targets_min": 4960 | |
| }, | |
| { | |
| "epoch": 5.098436595251882, | |
| "grad_norm": 0.46639560747858133, | |
| "learning_rate": 8.348198119843846e-06, | |
| "loss": 0.1637, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07922901213169098, | |
| "step": 4405, | |
| "valid_targets_mean": 6527.1, | |
| "valid_targets_min": 5399 | |
| }, | |
| { | |
| "epoch": 5.104226983207875, | |
| "grad_norm": 0.4974472656784201, | |
| "learning_rate": 8.301335441452486e-06, | |
| "loss": 0.1694, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0812421441078186, | |
| "step": 4410, | |
| "valid_targets_mean": 4995.0, | |
| "valid_targets_min": 618 | |
| }, | |
| { | |
| "epoch": 5.110017371163868, | |
| "grad_norm": 0.4384624621750658, | |
| "learning_rate": 8.25457019449016e-06, | |
| "loss": 0.1804, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08994653075933456, | |
| "step": 4415, | |
| "valid_targets_mean": 7219.4, | |
| "valid_targets_min": 5764 | |
| }, | |
| { | |
| "epoch": 5.115807759119861, | |
| "grad_norm": 0.4506617842709948, | |
| "learning_rate": 8.207902768437643e-06, | |
| "loss": 0.1483, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08390418440103531, | |
| "step": 4420, | |
| "valid_targets_mean": 7998.5, | |
| "valid_targets_min": 5902 | |
| }, | |
| { | |
| "epoch": 5.121598147075854, | |
| "grad_norm": 0.4223545408283372, | |
| "learning_rate": 8.161333551960993e-06, | |
| "loss": 0.1563, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.06642226874828339, | |
| "step": 4425, | |
| "valid_targets_mean": 7377.4, | |
| "valid_targets_min": 5015 | |
| }, | |
| { | |
| "epoch": 5.127388535031847, | |
| "grad_norm": 0.4166683550696679, | |
| "learning_rate": 8.114862932908356e-06, | |
| "loss": 0.1631, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0684872716665268, | |
| "step": 4430, | |
| "valid_targets_mean": 6794.0, | |
| "valid_targets_min": 5283 | |
| }, | |
| { | |
| "epoch": 5.13317892298784, | |
| "grad_norm": 0.44850588452073403, | |
| "learning_rate": 8.068491298306718e-06, | |
| "loss": 0.1691, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0934000313282013, | |
| "step": 4435, | |
| "valid_targets_mean": 6656.4, | |
| "valid_targets_min": 4364 | |
| }, | |
| { | |
| "epoch": 5.138969310943834, | |
| "grad_norm": 0.4045204129415732, | |
| "learning_rate": 8.022219034358676e-06, | |
| "loss": 0.1673, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07664947211742401, | |
| "step": 4440, | |
| "valid_targets_mean": 7178.4, | |
| "valid_targets_min": 4448 | |
| }, | |
| { | |
| "epoch": 5.1447596988998265, | |
| "grad_norm": 0.4792195984467097, | |
| "learning_rate": 7.976046526439215e-06, | |
| "loss": 0.1659, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09860777854919434, | |
| "step": 4445, | |
| "valid_targets_mean": 7619.8, | |
| "valid_targets_min": 5737 | |
| }, | |
| { | |
| "epoch": 5.150550086855819, | |
| "grad_norm": 0.38924917775303713, | |
| "learning_rate": 7.929974159092531e-06, | |
| "loss": 0.1471, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.06915652006864548, | |
| "step": 4450, | |
| "valid_targets_mean": 8754.4, | |
| "valid_targets_min": 5752 | |
| }, | |
| { | |
| "epoch": 5.156340474811812, | |
| "grad_norm": 0.4842231172925656, | |
| "learning_rate": 7.884002316028787e-06, | |
| "loss": 0.1483, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0772770345211029, | |
| "step": 4455, | |
| "valid_targets_mean": 6341.0, | |
| "valid_targets_min": 4543 | |
| }, | |
| { | |
| "epoch": 5.162130862767805, | |
| "grad_norm": 0.43535274367025806, | |
| "learning_rate": 7.838131380120962e-06, | |
| "loss": 0.144, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08321338891983032, | |
| "step": 4460, | |
| "valid_targets_mean": 7181.9, | |
| "valid_targets_min": 4563 | |
| }, | |
| { | |
| "epoch": 5.167921250723799, | |
| "grad_norm": 0.7920145323587261, | |
| "learning_rate": 7.792361733401599e-06, | |
| "loss": 0.1683, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.05222175642848015, | |
| "step": 4465, | |
| "valid_targets_mean": 1104.4, | |
| "valid_targets_min": 161 | |
| }, | |
| { | |
| "epoch": 5.173711638679792, | |
| "grad_norm": 0.44403128554587973, | |
| "learning_rate": 7.7466937570597e-06, | |
| "loss": 0.1527, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07515177875757217, | |
| "step": 4470, | |
| "valid_targets_mean": 7018.2, | |
| "valid_targets_min": 5424 | |
| }, | |
| { | |
| "epoch": 5.1795020266357845, | |
| "grad_norm": 0.4722726033181905, | |
| "learning_rate": 7.701127831437499e-06, | |
| "loss": 0.1733, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08935241401195526, | |
| "step": 4475, | |
| "valid_targets_mean": 7738.4, | |
| "valid_targets_min": 5329 | |
| }, | |
| { | |
| "epoch": 5.185292414591777, | |
| "grad_norm": 0.45611397186572844, | |
| "learning_rate": 7.655664336027296e-06, | |
| "loss": 0.1782, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08397034555673599, | |
| "step": 4480, | |
| "valid_targets_mean": 6426.6, | |
| "valid_targets_min": 5560 | |
| }, | |
| { | |
| "epoch": 5.191082802547771, | |
| "grad_norm": 0.49322397234122534, | |
| "learning_rate": 7.610303649468323e-06, | |
| "loss": 0.1723, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08556440472602844, | |
| "step": 4485, | |
| "valid_targets_mean": 6169.9, | |
| "valid_targets_min": 3446 | |
| }, | |
| { | |
| "epoch": 5.196873190503764, | |
| "grad_norm": 0.4306990762052569, | |
| "learning_rate": 7.5650461495435775e-06, | |
| "loss": 0.167, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08588321506977081, | |
| "step": 4490, | |
| "valid_targets_mean": 7204.6, | |
| "valid_targets_min": 5381 | |
| }, | |
| { | |
| "epoch": 5.202663578459757, | |
| "grad_norm": 0.42431598510637303, | |
| "learning_rate": 7.519892213176669e-06, | |
| "loss": 0.1705, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0885540097951889, | |
| "step": 4495, | |
| "valid_targets_mean": 7600.4, | |
| "valid_targets_min": 5241 | |
| }, | |
| { | |
| "epoch": 5.20845396641575, | |
| "grad_norm": 0.42897458180789355, | |
| "learning_rate": 7.474842216428695e-06, | |
| "loss": 0.1666, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08960206806659698, | |
| "step": 4500, | |
| "valid_targets_mean": 7098.5, | |
| "valid_targets_min": 5607 | |
| }, | |
| { | |
| "epoch": 5.2142443543717425, | |
| "grad_norm": 0.4657738884299629, | |
| "learning_rate": 7.429896534495076e-06, | |
| "loss": 0.1696, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09080995619297028, | |
| "step": 4505, | |
| "valid_targets_mean": 6955.1, | |
| "valid_targets_min": 5530 | |
| }, | |
| { | |
| "epoch": 5.220034742327736, | |
| "grad_norm": 0.46053355463201484, | |
| "learning_rate": 7.385055541702489e-06, | |
| "loss": 0.171, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08669829368591309, | |
| "step": 4510, | |
| "valid_targets_mean": 6748.1, | |
| "valid_targets_min": 5364 | |
| }, | |
| { | |
| "epoch": 5.225825130283729, | |
| "grad_norm": 0.4683200226871999, | |
| "learning_rate": 7.340319611505682e-06, | |
| "loss": 0.1706, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0833309143781662, | |
| "step": 4515, | |
| "valid_targets_mean": 6273.9, | |
| "valid_targets_min": 5438 | |
| }, | |
| { | |
| "epoch": 5.231615518239722, | |
| "grad_norm": 0.42873974780150437, | |
| "learning_rate": 7.295689116484419e-06, | |
| "loss": 0.1659, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07850541174411774, | |
| "step": 4520, | |
| "valid_targets_mean": 6419.5, | |
| "valid_targets_min": 4879 | |
| }, | |
| { | |
| "epoch": 5.237405906195715, | |
| "grad_norm": 0.43734828472291104, | |
| "learning_rate": 7.251164428340349e-06, | |
| "loss": 0.1656, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07928910851478577, | |
| "step": 4525, | |
| "valid_targets_mean": 6483.4, | |
| "valid_targets_min": 4309 | |
| }, | |
| { | |
| "epoch": 5.2431962941517085, | |
| "grad_norm": 0.4825423523548664, | |
| "learning_rate": 7.206745917893918e-06, | |
| "loss": 0.138, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09682394564151764, | |
| "step": 4530, | |
| "valid_targets_mean": 6825.0, | |
| "valid_targets_min": 5205 | |
| }, | |
| { | |
| "epoch": 5.248986682107701, | |
| "grad_norm": 0.49377635032991485, | |
| "learning_rate": 7.162433955081283e-06, | |
| "loss": 0.166, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07656551897525787, | |
| "step": 4535, | |
| "valid_targets_mean": 6729.4, | |
| "valid_targets_min": 5200 | |
| }, | |
| { | |
| "epoch": 5.254777070063694, | |
| "grad_norm": 0.4554574464975237, | |
| "learning_rate": 7.1182289089512105e-06, | |
| "loss": 0.163, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07981318235397339, | |
| "step": 4540, | |
| "valid_targets_mean": 6094.0, | |
| "valid_targets_min": 5302 | |
| }, | |
| { | |
| "epoch": 5.260567458019687, | |
| "grad_norm": 0.457518491044438, | |
| "learning_rate": 7.074131147662042e-06, | |
| "loss": 0.1684, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08653154969215393, | |
| "step": 4545, | |
| "valid_targets_mean": 7140.0, | |
| "valid_targets_min": 5270 | |
| }, | |
| { | |
| "epoch": 5.26635784597568, | |
| "grad_norm": 0.4011706541140575, | |
| "learning_rate": 7.030141038478584e-06, | |
| "loss": 0.1606, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07779765874147415, | |
| "step": 4550, | |
| "valid_targets_mean": 7425.0, | |
| "valid_targets_min": 5281 | |
| }, | |
| { | |
| "epoch": 5.272148233931674, | |
| "grad_norm": 0.4491415779590304, | |
| "learning_rate": 6.986258947769087e-06, | |
| "loss": 0.1691, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10509377717971802, | |
| "step": 4555, | |
| "valid_targets_mean": 7063.4, | |
| "valid_targets_min": 5386 | |
| }, | |
| { | |
| "epoch": 5.2779386218876665, | |
| "grad_norm": 0.4384591648317973, | |
| "learning_rate": 6.942485241002173e-06, | |
| "loss": 0.1669, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08581419289112091, | |
| "step": 4560, | |
| "valid_targets_mean": 6827.0, | |
| "valid_targets_min": 4547 | |
| }, | |
| { | |
| "epoch": 5.283729009843659, | |
| "grad_norm": 0.603270492629835, | |
| "learning_rate": 6.898820282743792e-06, | |
| "loss": 0.144, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.03785271942615509, | |
| "step": 4565, | |
| "valid_targets_mean": 2047.0, | |
| "valid_targets_min": 165 | |
| }, | |
| { | |
| "epoch": 5.289519397799652, | |
| "grad_norm": 0.4338840236261938, | |
| "learning_rate": 6.855264436654199e-06, | |
| "loss": 0.1493, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07795995473861694, | |
| "step": 4570, | |
| "valid_targets_mean": 6754.1, | |
| "valid_targets_min": 5196 | |
| }, | |
| { | |
| "epoch": 5.295309785755646, | |
| "grad_norm": 0.49540523131501424, | |
| "learning_rate": 6.811818065484918e-06, | |
| "loss": 0.1642, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08165041357278824, | |
| "step": 4575, | |
| "valid_targets_mean": 5985.6, | |
| "valid_targets_min": 5028 | |
| }, | |
| { | |
| "epoch": 5.301100173711639, | |
| "grad_norm": 0.4669333045816074, | |
| "learning_rate": 6.7684815310756965e-06, | |
| "loss": 0.1674, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07661072909832001, | |
| "step": 4580, | |
| "valid_targets_mean": 5963.9, | |
| "valid_targets_min": 5237 | |
| }, | |
| { | |
| "epoch": 5.306890561667632, | |
| "grad_norm": 0.4820832943223326, | |
| "learning_rate": 6.72525519435155e-06, | |
| "loss": 0.1619, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0846099928021431, | |
| "step": 4585, | |
| "valid_targets_mean": 6319.0, | |
| "valid_targets_min": 5412 | |
| }, | |
| { | |
| "epoch": 5.3126809496236245, | |
| "grad_norm": 0.42051650052749845, | |
| "learning_rate": 6.682139415319688e-06, | |
| "loss": 0.1541, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.06711379438638687, | |
| "step": 4590, | |
| "valid_targets_mean": 7125.8, | |
| "valid_targets_min": 5080 | |
| }, | |
| { | |
| "epoch": 5.318471337579618, | |
| "grad_norm": 0.4652762288890063, | |
| "learning_rate": 6.639134553066568e-06, | |
| "loss": 0.1633, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07977567613124847, | |
| "step": 4595, | |
| "valid_targets_mean": 6123.5, | |
| "valid_targets_min": 4521 | |
| }, | |
| { | |
| "epoch": 5.324261725535611, | |
| "grad_norm": 0.6454278429206071, | |
| "learning_rate": 6.5962409657548875e-06, | |
| "loss": 0.1913, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12123078107833862, | |
| "step": 4600, | |
| "valid_targets_mean": 6020.6, | |
| "valid_targets_min": 2493 | |
| }, | |
| { | |
| "epoch": 5.330052113491604, | |
| "grad_norm": 0.8629500695343556, | |
| "learning_rate": 6.5534590106205865e-06, | |
| "loss": 0.2119, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08393067121505737, | |
| "step": 4605, | |
| "valid_targets_mean": 3953.5, | |
| "valid_targets_min": 1177 | |
| }, | |
| { | |
| "epoch": 5.335842501447597, | |
| "grad_norm": 0.6295404053425543, | |
| "learning_rate": 6.5107890439698965e-06, | |
| "loss": 0.2207, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13039299845695496, | |
| "step": 4610, | |
| "valid_targets_mean": 4950.8, | |
| "valid_targets_min": 2799 | |
| }, | |
| { | |
| "epoch": 5.34163288940359, | |
| "grad_norm": 0.6278474637692992, | |
| "learning_rate": 6.468231421176359e-06, | |
| "loss": 0.2073, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11538614332675934, | |
| "step": 4615, | |
| "valid_targets_mean": 5312.0, | |
| "valid_targets_min": 2973 | |
| }, | |
| { | |
| "epoch": 5.347423277359583, | |
| "grad_norm": 0.6286400950974823, | |
| "learning_rate": 6.425786496677855e-06, | |
| "loss": 0.2084, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09965649247169495, | |
| "step": 4620, | |
| "valid_targets_mean": 3775.5, | |
| "valid_targets_min": 980 | |
| }, | |
| { | |
| "epoch": 5.353213665315576, | |
| "grad_norm": 0.5945486249462685, | |
| "learning_rate": 6.3834546239736925e-06, | |
| "loss": 0.2228, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11746525764465332, | |
| "step": 4625, | |
| "valid_targets_mean": 5055.4, | |
| "valid_targets_min": 1512 | |
| }, | |
| { | |
| "epoch": 5.359004053271569, | |
| "grad_norm": 0.656038142968123, | |
| "learning_rate": 6.3412361556216065e-06, | |
| "loss": 0.2028, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.079439178109169, | |
| "step": 4630, | |
| "valid_targets_mean": 3600.8, | |
| "valid_targets_min": 1848 | |
| }, | |
| { | |
| "epoch": 5.364794441227562, | |
| "grad_norm": 0.5666988911136864, | |
| "learning_rate": 6.2991314432348714e-06, | |
| "loss": 0.2107, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07719736546278, | |
| "step": 4635, | |
| "valid_targets_mean": 3369.8, | |
| "valid_targets_min": 1381 | |
| }, | |
| { | |
| "epoch": 5.370584829183556, | |
| "grad_norm": 0.5244894286826159, | |
| "learning_rate": 6.257140837479352e-06, | |
| "loss": 0.2078, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11062712222337723, | |
| "step": 4640, | |
| "valid_targets_mean": 6283.2, | |
| "valid_targets_min": 3600 | |
| }, | |
| { | |
| "epoch": 5.3763752171395485, | |
| "grad_norm": 0.6203824635854146, | |
| "learning_rate": 6.2152646880705835e-06, | |
| "loss": 0.2229, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10925555974245071, | |
| "step": 4645, | |
| "valid_targets_mean": 5242.5, | |
| "valid_targets_min": 2540 | |
| }, | |
| { | |
| "epoch": 5.382165605095541, | |
| "grad_norm": 0.5713876724097097, | |
| "learning_rate": 6.173503343770864e-06, | |
| "loss": 0.2065, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09428058564662933, | |
| "step": 4650, | |
| "valid_targets_mean": 6246.5, | |
| "valid_targets_min": 2157 | |
| }, | |
| { | |
| "epoch": 5.387955993051534, | |
| "grad_norm": 0.6140290374368306, | |
| "learning_rate": 6.131857152386329e-06, | |
| "loss": 0.2097, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10542860627174377, | |
| "step": 4655, | |
| "valid_targets_mean": 4556.5, | |
| "valid_targets_min": 1720 | |
| }, | |
| { | |
| "epoch": 5.393746381007528, | |
| "grad_norm": 0.6065323078628322, | |
| "learning_rate": 6.0903264607640955e-06, | |
| "loss": 0.2158, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1147131621837616, | |
| "step": 4660, | |
| "valid_targets_mean": 5116.5, | |
| "valid_targets_min": 1982 | |
| }, | |
| { | |
| "epoch": 5.399536768963521, | |
| "grad_norm": 0.5810831150496794, | |
| "learning_rate": 6.048911614789325e-06, | |
| "loss": 0.2094, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10943673551082611, | |
| "step": 4665, | |
| "valid_targets_mean": 4464.5, | |
| "valid_targets_min": 2718 | |
| }, | |
| { | |
| "epoch": 5.405327156919514, | |
| "grad_norm": 0.6448830357345737, | |
| "learning_rate": 6.007612959382392e-06, | |
| "loss": 0.205, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1034761518239975, | |
| "step": 4670, | |
| "valid_targets_mean": 3418.5, | |
| "valid_targets_min": 1451 | |
| }, | |
| { | |
| "epoch": 5.4111175448755064, | |
| "grad_norm": 0.6646340852402701, | |
| "learning_rate": 5.966430838495969e-06, | |
| "loss": 0.2025, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10305717587471008, | |
| "step": 4675, | |
| "valid_targets_mean": 3990.4, | |
| "valid_targets_min": 1861 | |
| }, | |
| { | |
| "epoch": 5.416907932831499, | |
| "grad_norm": 0.6787610049432943, | |
| "learning_rate": 5.925365595112189e-06, | |
| "loss": 0.1992, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11291109025478363, | |
| "step": 4680, | |
| "valid_targets_mean": 4065.4, | |
| "valid_targets_min": 2100 | |
| }, | |
| { | |
| "epoch": 5.422698320787493, | |
| "grad_norm": 0.6468199469370208, | |
| "learning_rate": 5.8844175712397826e-06, | |
| "loss": 0.2043, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11317706108093262, | |
| "step": 4685, | |
| "valid_targets_mean": 4132.4, | |
| "valid_targets_min": 3234 | |
| }, | |
| { | |
| "epoch": 5.428488708743486, | |
| "grad_norm": 0.7000979053549637, | |
| "learning_rate": 5.843587107911209e-06, | |
| "loss": 0.214, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12522569298744202, | |
| "step": 4690, | |
| "valid_targets_mean": 4246.2, | |
| "valid_targets_min": 2650 | |
| }, | |
| { | |
| "epoch": 5.434279096699479, | |
| "grad_norm": 0.6580517542861027, | |
| "learning_rate": 5.802874545179848e-06, | |
| "loss": 0.2062, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0871807187795639, | |
| "step": 4695, | |
| "valid_targets_mean": 4096.5, | |
| "valid_targets_min": 1719 | |
| }, | |
| { | |
| "epoch": 5.440069484655472, | |
| "grad_norm": 0.6577612345149414, | |
| "learning_rate": 5.76228022211716e-06, | |
| "loss": 0.2119, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10219301283359528, | |
| "step": 4700, | |
| "valid_targets_mean": 3933.1, | |
| "valid_targets_min": 1020 | |
| }, | |
| { | |
| "epoch": 5.445859872611465, | |
| "grad_norm": 0.659954032330438, | |
| "learning_rate": 5.721804476809836e-06, | |
| "loss": 0.1956, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10725036263465881, | |
| "step": 4705, | |
| "valid_targets_mean": 3551.1, | |
| "valid_targets_min": 2130 | |
| }, | |
| { | |
| "epoch": 5.451650260567458, | |
| "grad_norm": 0.6312457586493002, | |
| "learning_rate": 5.681447646357012e-06, | |
| "loss": 0.1927, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07873150706291199, | |
| "step": 4710, | |
| "valid_targets_mean": 3535.0, | |
| "valid_targets_min": 865 | |
| }, | |
| { | |
| "epoch": 5.457440648523451, | |
| "grad_norm": 0.7722552413609229, | |
| "learning_rate": 5.641210066867455e-06, | |
| "loss": 0.1963, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1350419670343399, | |
| "step": 4715, | |
| "valid_targets_mean": 3716.5, | |
| "valid_targets_min": 3144 | |
| }, | |
| { | |
| "epoch": 5.463231036479444, | |
| "grad_norm": 0.7467110819892256, | |
| "learning_rate": 5.601092073456757e-06, | |
| "loss": 0.206, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09063935279846191, | |
| "step": 4720, | |
| "valid_targets_mean": 3262.2, | |
| "valid_targets_min": 2050 | |
| }, | |
| { | |
| "epoch": 5.469021424435438, | |
| "grad_norm": 0.6632300609186226, | |
| "learning_rate": 5.561094000244534e-06, | |
| "loss": 0.204, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10219931602478027, | |
| "step": 4725, | |
| "valid_targets_mean": 3296.5, | |
| "valid_targets_min": 1570 | |
| }, | |
| { | |
| "epoch": 5.47481181239143, | |
| "grad_norm": 0.6338761495132271, | |
| "learning_rate": 5.52121618035167e-06, | |
| "loss": 0.1931, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10076398402452469, | |
| "step": 4730, | |
| "valid_targets_mean": 4806.4, | |
| "valid_targets_min": 1795 | |
| }, | |
| { | |
| "epoch": 5.480602200347423, | |
| "grad_norm": 0.58715133275694, | |
| "learning_rate": 5.481458945897524e-06, | |
| "loss": 0.1968, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09306555986404419, | |
| "step": 4735, | |
| "valid_targets_mean": 4342.4, | |
| "valid_targets_min": 1934 | |
| }, | |
| { | |
| "epoch": 5.486392588303416, | |
| "grad_norm": 0.678016304337756, | |
| "learning_rate": 5.44182262799718e-06, | |
| "loss": 0.1979, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1021447628736496, | |
| "step": 4740, | |
| "valid_targets_mean": 3675.4, | |
| "valid_targets_min": 2557 | |
| }, | |
| { | |
| "epoch": 5.492182976259409, | |
| "grad_norm": 0.6811772712489489, | |
| "learning_rate": 5.402307556758648e-06, | |
| "loss": 0.1954, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11167696118354797, | |
| "step": 4745, | |
| "valid_targets_mean": 4550.4, | |
| "valid_targets_min": 2663 | |
| }, | |
| { | |
| "epoch": 5.497973364215403, | |
| "grad_norm": 0.6371050216182214, | |
| "learning_rate": 5.362914061280173e-06, | |
| "loss": 0.191, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08882875740528107, | |
| "step": 4750, | |
| "valid_targets_mean": 4604.0, | |
| "valid_targets_min": 2016 | |
| }, | |
| { | |
| "epoch": 5.503763752171396, | |
| "grad_norm": 0.7119686877592675, | |
| "learning_rate": 5.323642469647457e-06, | |
| "loss": 0.1971, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08872997760772705, | |
| "step": 4755, | |
| "valid_targets_mean": 2976.8, | |
| "valid_targets_min": 1367 | |
| }, | |
| { | |
| "epoch": 5.509554140127388, | |
| "grad_norm": 0.6917916761818202, | |
| "learning_rate": 5.284493108930935e-06, | |
| "loss": 0.1996, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09722135961055756, | |
| "step": 4760, | |
| "valid_targets_mean": 3542.5, | |
| "valid_targets_min": 2026 | |
| }, | |
| { | |
| "epoch": 5.515344528083381, | |
| "grad_norm": 0.6799912825803575, | |
| "learning_rate": 5.245466305183042e-06, | |
| "loss": 0.1874, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08388345688581467, | |
| "step": 4765, | |
| "valid_targets_mean": 3247.2, | |
| "valid_targets_min": 2033 | |
| }, | |
| { | |
| "epoch": 5.521134916039375, | |
| "grad_norm": 0.8200711667381476, | |
| "learning_rate": 5.2065623834355205e-06, | |
| "loss": 0.1956, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09685410559177399, | |
| "step": 4770, | |
| "valid_targets_mean": 2201.6, | |
| "valid_targets_min": 1283 | |
| }, | |
| { | |
| "epoch": 5.526925303995368, | |
| "grad_norm": 0.773411446394524, | |
| "learning_rate": 5.167781667696692e-06, | |
| "loss": 0.1967, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10723498463630676, | |
| "step": 4775, | |
| "valid_targets_mean": 2824.8, | |
| "valid_targets_min": 1636 | |
| }, | |
| { | |
| "epoch": 5.532715691951361, | |
| "grad_norm": 0.7105868546737427, | |
| "learning_rate": 5.129124480948775e-06, | |
| "loss": 0.199, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10746259987354279, | |
| "step": 4780, | |
| "valid_targets_mean": 3844.4, | |
| "valid_targets_min": 1629 | |
| }, | |
| { | |
| "epoch": 5.5385060799073536, | |
| "grad_norm": 0.7119922314320859, | |
| "learning_rate": 5.090591145145169e-06, | |
| "loss": 0.1936, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09051588177680969, | |
| "step": 4785, | |
| "valid_targets_mean": 3048.6, | |
| "valid_targets_min": 2080 | |
| }, | |
| { | |
| "epoch": 5.544296467863347, | |
| "grad_norm": 0.7116798621330098, | |
| "learning_rate": 5.0521819812078046e-06, | |
| "loss": 0.1982, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09174008667469025, | |
| "step": 4790, | |
| "valid_targets_mean": 3111.8, | |
| "valid_targets_min": 1153 | |
| }, | |
| { | |
| "epoch": 5.55008685581934, | |
| "grad_norm": 0.6706055056945278, | |
| "learning_rate": 5.013897309024462e-06, | |
| "loss": 0.1935, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10536851733922958, | |
| "step": 4795, | |
| "valid_targets_mean": 4069.0, | |
| "valid_targets_min": 1858 | |
| }, | |
| { | |
| "epoch": 5.555877243775333, | |
| "grad_norm": 0.6242996097491735, | |
| "learning_rate": 4.975737447446076e-06, | |
| "loss": 0.2014, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08580464869737625, | |
| "step": 4800, | |
| "valid_targets_mean": 3415.2, | |
| "valid_targets_min": 1639 | |
| }, | |
| { | |
| "epoch": 5.561667631731326, | |
| "grad_norm": 0.7034187772300746, | |
| "learning_rate": 4.937702714284132e-06, | |
| "loss": 0.1919, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07671986520290375, | |
| "step": 4805, | |
| "valid_targets_mean": 3267.6, | |
| "valid_targets_min": 1394 | |
| }, | |
| { | |
| "epoch": 5.567458019687319, | |
| "grad_norm": 0.6517323852977768, | |
| "learning_rate": 4.899793426307982e-06, | |
| "loss": 0.1973, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10675493627786636, | |
| "step": 4810, | |
| "valid_targets_mean": 4775.9, | |
| "valid_targets_min": 1844 | |
| }, | |
| { | |
| "epoch": 5.573248407643312, | |
| "grad_norm": 0.7018212376921986, | |
| "learning_rate": 4.862009899242219e-06, | |
| "loss": 0.1933, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08690626919269562, | |
| "step": 4815, | |
| "valid_targets_mean": 3317.1, | |
| "valid_targets_min": 1305 | |
| }, | |
| { | |
| "epoch": 5.579038795599305, | |
| "grad_norm": 0.69114193929696, | |
| "learning_rate": 4.824352447764049e-06, | |
| "loss": 0.1969, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1058984249830246, | |
| "step": 4820, | |
| "valid_targets_mean": 4290.8, | |
| "valid_targets_min": 2185 | |
| }, | |
| { | |
| "epoch": 5.584829183555298, | |
| "grad_norm": 0.7040668680153952, | |
| "learning_rate": 4.786821385500653e-06, | |
| "loss": 0.1914, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1182374507188797, | |
| "step": 4825, | |
| "valid_targets_mean": 4024.9, | |
| "valid_targets_min": 2349 | |
| }, | |
| { | |
| "epoch": 5.590619571511291, | |
| "grad_norm": 0.6527180269033891, | |
| "learning_rate": 4.749417025026615e-06, | |
| "loss": 0.1925, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08381571620702744, | |
| "step": 4830, | |
| "valid_targets_mean": 3453.0, | |
| "valid_targets_min": 1607 | |
| }, | |
| { | |
| "epoch": 5.596409959467284, | |
| "grad_norm": 0.7116519649417133, | |
| "learning_rate": 4.7121396778612606e-06, | |
| "loss": 0.1998, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10524024069309235, | |
| "step": 4835, | |
| "valid_targets_mean": 3389.8, | |
| "valid_targets_min": 1507 | |
| }, | |
| { | |
| "epoch": 5.6022003474232775, | |
| "grad_norm": 0.65985189052359, | |
| "learning_rate": 4.67498965446612e-06, | |
| "loss": 0.1913, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08167605847120285, | |
| "step": 4840, | |
| "valid_targets_mean": 3773.9, | |
| "valid_targets_min": 1317 | |
| }, | |
| { | |
| "epoch": 5.60799073537927, | |
| "grad_norm": 0.6503048626054666, | |
| "learning_rate": 4.637967264242309e-06, | |
| "loss": 0.1964, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09242589771747589, | |
| "step": 4845, | |
| "valid_targets_mean": 3881.0, | |
| "valid_targets_min": 3355 | |
| }, | |
| { | |
| "epoch": 5.613781123335263, | |
| "grad_norm": 0.6594418766777953, | |
| "learning_rate": 4.601072815527961e-06, | |
| "loss": 0.2025, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.099497951567173, | |
| "step": 4850, | |
| "valid_targets_mean": 3951.9, | |
| "valid_targets_min": 2999 | |
| }, | |
| { | |
| "epoch": 5.619571511291257, | |
| "grad_norm": 0.6860624635409228, | |
| "learning_rate": 4.564306615595657e-06, | |
| "loss": 0.1958, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08808408677577972, | |
| "step": 4855, | |
| "valid_targets_mean": 4086.5, | |
| "valid_targets_min": 1739 | |
| }, | |
| { | |
| "epoch": 5.62536189924725, | |
| "grad_norm": 0.6535359815334453, | |
| "learning_rate": 4.527668970649875e-06, | |
| "loss": 0.1936, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08575412631034851, | |
| "step": 4860, | |
| "valid_targets_mean": 3298.0, | |
| "valid_targets_min": 1140 | |
| }, | |
| { | |
| "epoch": 5.631152287203243, | |
| "grad_norm": 0.6448606842611453, | |
| "learning_rate": 4.491160185824419e-06, | |
| "loss": 0.1825, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0939522311091423, | |
| "step": 4865, | |
| "valid_targets_mean": 4012.1, | |
| "valid_targets_min": 2779 | |
| }, | |
| { | |
| "epoch": 5.6369426751592355, | |
| "grad_norm": 0.6641592030702549, | |
| "learning_rate": 4.454780565179894e-06, | |
| "loss": 0.1939, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0876445323228836, | |
| "step": 4870, | |
| "valid_targets_mean": 3636.9, | |
| "valid_targets_min": 1183 | |
| }, | |
| { | |
| "epoch": 5.642733063115228, | |
| "grad_norm": 0.6922167993474277, | |
| "learning_rate": 4.418530411701183e-06, | |
| "loss": 0.1841, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11794327944517136, | |
| "step": 4875, | |
| "valid_targets_mean": 4269.0, | |
| "valid_targets_min": 3041 | |
| }, | |
| { | |
| "epoch": 5.648523451071222, | |
| "grad_norm": 0.6710735679921558, | |
| "learning_rate": 4.382410027294901e-06, | |
| "loss": 0.1996, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0949588268995285, | |
| "step": 4880, | |
| "valid_targets_mean": 3350.1, | |
| "valid_targets_min": 1726 | |
| }, | |
| { | |
| "epoch": 5.654313839027215, | |
| "grad_norm": 0.5809522620469011, | |
| "learning_rate": 4.346419712786898e-06, | |
| "loss": 0.1894, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08865835517644882, | |
| "step": 4885, | |
| "valid_targets_mean": 4747.4, | |
| "valid_targets_min": 1743 | |
| }, | |
| { | |
| "epoch": 5.660104226983208, | |
| "grad_norm": 0.6383510256052882, | |
| "learning_rate": 4.3105597679197415e-06, | |
| "loss": 0.1981, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09135545790195465, | |
| "step": 4890, | |
| "valid_targets_mean": 3702.8, | |
| "valid_targets_min": 1234 | |
| }, | |
| { | |
| "epoch": 5.665894614939201, | |
| "grad_norm": 0.6754634424721645, | |
| "learning_rate": 4.274830491350226e-06, | |
| "loss": 0.2002, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10354183614253998, | |
| "step": 4895, | |
| "valid_targets_mean": 4079.0, | |
| "valid_targets_min": 2303 | |
| }, | |
| { | |
| "epoch": 5.6716850028951935, | |
| "grad_norm": 0.7095886049137189, | |
| "learning_rate": 4.239232180646895e-06, | |
| "loss": 0.2013, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09504668414592743, | |
| "step": 4900, | |
| "valid_targets_mean": 3830.8, | |
| "valid_targets_min": 1651 | |
| }, | |
| { | |
| "epoch": 5.677475390851187, | |
| "grad_norm": 0.6960742471213217, | |
| "learning_rate": 4.2037651322875385e-06, | |
| "loss": 0.1926, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09748727083206177, | |
| "step": 4905, | |
| "valid_targets_mean": 2876.2, | |
| "valid_targets_min": 1374 | |
| }, | |
| { | |
| "epoch": 5.68326577880718, | |
| "grad_norm": 0.7341851009140241, | |
| "learning_rate": 4.168429641656735e-06, | |
| "loss": 0.1872, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09375683963298798, | |
| "step": 4910, | |
| "valid_targets_mean": 3467.5, | |
| "valid_targets_min": 1625 | |
| }, | |
| { | |
| "epoch": 5.689056166763173, | |
| "grad_norm": 0.6683721274985364, | |
| "learning_rate": 4.133226003043414e-06, | |
| "loss": 0.1839, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09662937372922897, | |
| "step": 4915, | |
| "valid_targets_mean": 4089.5, | |
| "valid_targets_min": 997 | |
| }, | |
| { | |
| "epoch": 5.694846554719167, | |
| "grad_norm": 0.6481627696138452, | |
| "learning_rate": 4.098154509638372e-06, | |
| "loss": 0.1831, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0757637545466423, | |
| "step": 4920, | |
| "valid_targets_mean": 3618.6, | |
| "valid_targets_min": 1179 | |
| }, | |
| { | |
| "epoch": 5.7006369426751595, | |
| "grad_norm": 0.7512215381813111, | |
| "learning_rate": 4.06321545353185e-06, | |
| "loss": 0.1859, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09299871325492859, | |
| "step": 4925, | |
| "valid_targets_mean": 3079.1, | |
| "valid_targets_min": 1444 | |
| }, | |
| { | |
| "epoch": 5.706427330631152, | |
| "grad_norm": 0.6808994582603312, | |
| "learning_rate": 4.02840912571109e-06, | |
| "loss": 0.1718, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07229427993297577, | |
| "step": 4930, | |
| "valid_targets_mean": 3486.8, | |
| "valid_targets_min": 1504 | |
| }, | |
| { | |
| "epoch": 5.712217718587145, | |
| "grad_norm": 0.6452902106629557, | |
| "learning_rate": 3.993735816057924e-06, | |
| "loss": 0.1779, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08849136531352997, | |
| "step": 4935, | |
| "valid_targets_mean": 3998.4, | |
| "valid_targets_min": 2302 | |
| }, | |
| { | |
| "epoch": 5.718008106543138, | |
| "grad_norm": 0.7621091499298214, | |
| "learning_rate": 3.959195813346342e-06, | |
| "loss": 0.1898, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11983069777488708, | |
| "step": 4940, | |
| "valid_targets_mean": 4282.5, | |
| "valid_targets_min": 1818 | |
| }, | |
| { | |
| "epoch": 5.723798494499132, | |
| "grad_norm": 0.7004998749919713, | |
| "learning_rate": 3.924789405240108e-06, | |
| "loss": 0.2066, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10091252624988556, | |
| "step": 4945, | |
| "valid_targets_mean": 4221.9, | |
| "valid_targets_min": 2633 | |
| }, | |
| { | |
| "epoch": 5.729588882455125, | |
| "grad_norm": 0.7052587606385302, | |
| "learning_rate": 3.89051687829034e-06, | |
| "loss": 0.1839, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09614616632461548, | |
| "step": 4950, | |
| "valid_targets_mean": 3522.4, | |
| "valid_targets_min": 2481 | |
| }, | |
| { | |
| "epoch": 5.7353792704111175, | |
| "grad_norm": 0.7015131244757292, | |
| "learning_rate": 3.856378517933152e-06, | |
| "loss": 0.1943, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10637455433607101, | |
| "step": 4955, | |
| "valid_targets_mean": 4058.8, | |
| "valid_targets_min": 2013 | |
| }, | |
| { | |
| "epoch": 5.74116965836711, | |
| "grad_norm": 0.7335595197985654, | |
| "learning_rate": 3.822374608487256e-06, | |
| "loss": 0.187, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10042333602905273, | |
| "step": 4960, | |
| "valid_targets_mean": 3723.5, | |
| "valid_targets_min": 1744 | |
| }, | |
| { | |
| "epoch": 5.746960046323103, | |
| "grad_norm": 0.6924352020727813, | |
| "learning_rate": 3.788505433151599e-06, | |
| "loss": 0.198, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10071320831775665, | |
| "step": 4965, | |
| "valid_targets_mean": 4180.9, | |
| "valid_targets_min": 1914 | |
| }, | |
| { | |
| "epoch": 5.752750434279097, | |
| "grad_norm": 0.7005163039353128, | |
| "learning_rate": 3.7547712740030085e-06, | |
| "loss": 0.1884, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08322249352931976, | |
| "step": 4970, | |
| "valid_targets_mean": 3668.6, | |
| "valid_targets_min": 3065 | |
| }, | |
| { | |
| "epoch": 5.75854082223509, | |
| "grad_norm": 0.6471601098849894, | |
| "learning_rate": 3.7211724119938297e-06, | |
| "loss": 0.1797, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10704165697097778, | |
| "step": 4975, | |
| "valid_targets_mean": 4473.2, | |
| "valid_targets_min": 2894 | |
| }, | |
| { | |
| "epoch": 5.764331210191083, | |
| "grad_norm": 0.732266303878351, | |
| "learning_rate": 3.68770912694961e-06, | |
| "loss": 0.1939, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09558729827404022, | |
| "step": 4980, | |
| "valid_targets_mean": 3395.6, | |
| "valid_targets_min": 2359 | |
| }, | |
| { | |
| "epoch": 5.7701215981470755, | |
| "grad_norm": 0.6915369674556678, | |
| "learning_rate": 3.654381697566749e-06, | |
| "loss": 0.1838, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08329612761735916, | |
| "step": 4985, | |
| "valid_targets_mean": 3114.5, | |
| "valid_targets_min": 1463 | |
| }, | |
| { | |
| "epoch": 5.775911986103069, | |
| "grad_norm": 0.7347954782252792, | |
| "learning_rate": 3.6211904014101752e-06, | |
| "loss": 0.1908, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10167864710092545, | |
| "step": 4990, | |
| "valid_targets_mean": 3588.1, | |
| "valid_targets_min": 1426 | |
| }, | |
| { | |
| "epoch": 5.781702374059062, | |
| "grad_norm": 0.693654917555153, | |
| "learning_rate": 3.5881355149110533e-06, | |
| "loss": 0.201, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09456135332584381, | |
| "step": 4995, | |
| "valid_targets_mean": 3405.5, | |
| "valid_targets_min": 1661 | |
| }, | |
| { | |
| "epoch": 5.787492762015055, | |
| "grad_norm": 0.6653941668059742, | |
| "learning_rate": 3.555217313364465e-06, | |
| "loss": 0.1943, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10260509699583054, | |
| "step": 5000, | |
| "valid_targets_mean": 4457.8, | |
| "valid_targets_min": 1529 | |
| }, | |
| { | |
| "epoch": 5.793283149971048, | |
| "grad_norm": 0.7980848239300522, | |
| "learning_rate": 3.522436070927129e-06, | |
| "loss": 0.2067, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11328806728124619, | |
| "step": 5005, | |
| "valid_targets_mean": 3272.9, | |
| "valid_targets_min": 1450 | |
| }, | |
| { | |
| "epoch": 5.7990735379270415, | |
| "grad_norm": 0.7329244914999775, | |
| "learning_rate": 3.4897920606150938e-06, | |
| "loss": 0.1894, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08768180012702942, | |
| "step": 5010, | |
| "valid_targets_mean": 3103.2, | |
| "valid_targets_min": 1625 | |
| }, | |
| { | |
| "epoch": 5.804863925883034, | |
| "grad_norm": 0.6504116880373527, | |
| "learning_rate": 3.457285554301497e-06, | |
| "loss": 0.2044, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08451417088508606, | |
| "step": 5015, | |
| "valid_targets_mean": 3952.2, | |
| "valid_targets_min": 1174 | |
| }, | |
| { | |
| "epoch": 5.810654313839027, | |
| "grad_norm": 0.7413148612675606, | |
| "learning_rate": 3.424916822714277e-06, | |
| "loss": 0.1987, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09813237190246582, | |
| "step": 5020, | |
| "valid_targets_mean": 4039.2, | |
| "valid_targets_min": 2882 | |
| }, | |
| { | |
| "epoch": 5.81644470179502, | |
| "grad_norm": 0.6950280348634762, | |
| "learning_rate": 3.3926861354339358e-06, | |
| "loss": 0.2019, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11226019263267517, | |
| "step": 5025, | |
| "valid_targets_mean": 4594.0, | |
| "valid_targets_min": 1946 | |
| }, | |
| { | |
| "epoch": 5.822235089751013, | |
| "grad_norm": 0.6906324382540923, | |
| "learning_rate": 3.3605937608912666e-06, | |
| "loss": 0.196, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09345898032188416, | |
| "step": 5030, | |
| "valid_targets_mean": 4082.6, | |
| "valid_targets_min": 2000 | |
| }, | |
| { | |
| "epoch": 5.828025477707007, | |
| "grad_norm": 0.7142374362982976, | |
| "learning_rate": 3.328639966365148e-06, | |
| "loss": 0.198, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10212359577417374, | |
| "step": 5035, | |
| "valid_targets_mean": 4596.1, | |
| "valid_targets_min": 2012 | |
| }, | |
| { | |
| "epoch": 5.8338158656629995, | |
| "grad_norm": 0.751407440395027, | |
| "learning_rate": 3.2968250179803097e-06, | |
| "loss": 0.1959, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09792028367519379, | |
| "step": 5040, | |
| "valid_targets_mean": 3094.4, | |
| "valid_targets_min": 719 | |
| }, | |
| { | |
| "epoch": 5.839606253618992, | |
| "grad_norm": 0.7400161140259937, | |
| "learning_rate": 3.265149180705096e-06, | |
| "loss": 0.1988, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09038306772708893, | |
| "step": 5045, | |
| "valid_targets_mean": 4496.2, | |
| "valid_targets_min": 3238 | |
| }, | |
| { | |
| "epoch": 5.845396641574985, | |
| "grad_norm": 0.7274957247678856, | |
| "learning_rate": 3.233612718349286e-06, | |
| "loss": 0.1868, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08311730623245239, | |
| "step": 5050, | |
| "valid_targets_mean": 3816.8, | |
| "valid_targets_min": 1372 | |
| }, | |
| { | |
| "epoch": 5.851187029530979, | |
| "grad_norm": 0.7813908012380086, | |
| "learning_rate": 3.202215893561884e-06, | |
| "loss": 0.1868, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10442608594894409, | |
| "step": 5055, | |
| "valid_targets_mean": 3306.1, | |
| "valid_targets_min": 2370 | |
| }, | |
| { | |
| "epoch": 5.856977417486972, | |
| "grad_norm": 0.6230264597997414, | |
| "learning_rate": 3.170958967828941e-06, | |
| "loss": 0.1944, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08658678084611893, | |
| "step": 5060, | |
| "valid_targets_mean": 2958.1, | |
| "valid_targets_min": 1576 | |
| }, | |
| { | |
| "epoch": 5.862767805442965, | |
| "grad_norm": 0.7117685524553994, | |
| "learning_rate": 3.13984220147135e-06, | |
| "loss": 0.1916, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10298854857683182, | |
| "step": 5065, | |
| "valid_targets_mean": 3758.5, | |
| "valid_targets_min": 1426 | |
| }, | |
| { | |
| "epoch": 5.8685581933989575, | |
| "grad_norm": 0.7330685461651261, | |
| "learning_rate": 3.108865853642715e-06, | |
| "loss": 0.2005, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08899010717868805, | |
| "step": 5070, | |
| "valid_targets_mean": 3117.0, | |
| "valid_targets_min": 1728 | |
| }, | |
| { | |
| "epoch": 5.874348581354951, | |
| "grad_norm": 0.766333671014308, | |
| "learning_rate": 3.078030182327174e-06, | |
| "loss": 0.1953, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1059906929731369, | |
| "step": 5075, | |
| "valid_targets_mean": 3264.2, | |
| "valid_targets_min": 1362 | |
| }, | |
| { | |
| "epoch": 5.880138969310944, | |
| "grad_norm": 0.6316516046760946, | |
| "learning_rate": 3.0473354443372337e-06, | |
| "loss": 0.1968, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09446334838867188, | |
| "step": 5080, | |
| "valid_targets_mean": 3836.0, | |
| "valid_targets_min": 852 | |
| }, | |
| { | |
| "epoch": 5.885929357266937, | |
| "grad_norm": 0.8278546280629885, | |
| "learning_rate": 3.016781895311671e-06, | |
| "loss": 0.1989, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12963464856147766, | |
| "step": 5085, | |
| "valid_targets_mean": 2764.1, | |
| "valid_targets_min": 1195 | |
| }, | |
| { | |
| "epoch": 5.89171974522293, | |
| "grad_norm": 0.6676519894325554, | |
| "learning_rate": 2.986369789713368e-06, | |
| "loss": 0.1947, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.104510597884655, | |
| "step": 5090, | |
| "valid_targets_mean": 4457.5, | |
| "valid_targets_min": 1564 | |
| }, | |
| { | |
| "epoch": 5.897510133178923, | |
| "grad_norm": 0.7287009986225004, | |
| "learning_rate": 2.956099380827213e-06, | |
| "loss": 0.197, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09116792678833008, | |
| "step": 5095, | |
| "valid_targets_mean": 4371.2, | |
| "valid_targets_min": 1975 | |
| }, | |
| { | |
| "epoch": 5.903300521134916, | |
| "grad_norm": 0.7364861255216155, | |
| "learning_rate": 2.9259709207579835e-06, | |
| "loss": 0.1822, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09816905856132507, | |
| "step": 5100, | |
| "valid_targets_mean": 3965.8, | |
| "valid_targets_min": 2426 | |
| }, | |
| { | |
| "epoch": 5.909090909090909, | |
| "grad_norm": 0.5730343476125825, | |
| "learning_rate": 2.895984660428235e-06, | |
| "loss": 0.1897, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09755247831344604, | |
| "step": 5105, | |
| "valid_targets_mean": 6384.6, | |
| "valid_targets_min": 2678 | |
| }, | |
| { | |
| "epoch": 5.914881297046902, | |
| "grad_norm": 0.6370394301993322, | |
| "learning_rate": 2.8661408495762423e-06, | |
| "loss": 0.1912, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0927160382270813, | |
| "step": 5110, | |
| "valid_targets_mean": 3543.2, | |
| "valid_targets_min": 2826 | |
| }, | |
| { | |
| "epoch": 5.920671685002895, | |
| "grad_norm": 0.6731370557606815, | |
| "learning_rate": 2.836439736753882e-06, | |
| "loss": 0.1885, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07997622340917587, | |
| "step": 5115, | |
| "valid_targets_mean": 3155.8, | |
| "valid_targets_min": 993 | |
| }, | |
| { | |
| "epoch": 5.926462072958889, | |
| "grad_norm": 0.6599899199310425, | |
| "learning_rate": 2.8068815693245976e-06, | |
| "loss": 0.1851, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10311765968799591, | |
| "step": 5120, | |
| "valid_targets_mean": 4166.5, | |
| "valid_targets_min": 2724 | |
| }, | |
| { | |
| "epoch": 5.9322524609148815, | |
| "grad_norm": 0.7027299860479045, | |
| "learning_rate": 2.777466593461313e-06, | |
| "loss": 0.1859, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09111426025629044, | |
| "step": 5125, | |
| "valid_targets_mean": 3079.5, | |
| "valid_targets_min": 1360 | |
| }, | |
| { | |
| "epoch": 5.938042848870874, | |
| "grad_norm": 0.6522068975482174, | |
| "learning_rate": 2.7481950541443957e-06, | |
| "loss": 0.1887, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09260795265436172, | |
| "step": 5130, | |
| "valid_targets_mean": 4114.9, | |
| "valid_targets_min": 2414 | |
| }, | |
| { | |
| "epoch": 5.943833236826867, | |
| "grad_norm": 0.7388334717337071, | |
| "learning_rate": 2.7190671951596124e-06, | |
| "loss": 0.2107, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11505642533302307, | |
| "step": 5135, | |
| "valid_targets_mean": 4524.8, | |
| "valid_targets_min": 2425 | |
| }, | |
| { | |
| "epoch": 5.949623624782861, | |
| "grad_norm": 0.8018275911931491, | |
| "learning_rate": 2.6900832590961033e-06, | |
| "loss": 0.1961, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09761355817317963, | |
| "step": 5140, | |
| "valid_targets_mean": 3035.2, | |
| "valid_targets_min": 1720 | |
| }, | |
| { | |
| "epoch": 5.955414012738854, | |
| "grad_norm": 0.7394931936417326, | |
| "learning_rate": 2.6612434873443425e-06, | |
| "loss": 0.191, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09674566984176636, | |
| "step": 5145, | |
| "valid_targets_mean": 3260.4, | |
| "valid_targets_min": 1621 | |
| }, | |
| { | |
| "epoch": 5.961204400694847, | |
| "grad_norm": 0.7282135820448314, | |
| "learning_rate": 2.632548120094167e-06, | |
| "loss": 0.1807, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09710457921028137, | |
| "step": 5150, | |
| "valid_targets_mean": 3214.6, | |
| "valid_targets_min": 1858 | |
| }, | |
| { | |
| "epoch": 5.966994788650839, | |
| "grad_norm": 0.7297952067172664, | |
| "learning_rate": 2.6039973963327245e-06, | |
| "loss": 0.1987, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09201544523239136, | |
| "step": 5155, | |
| "valid_targets_mean": 3254.2, | |
| "valid_targets_min": 1254 | |
| }, | |
| { | |
| "epoch": 5.972785176606832, | |
| "grad_norm": 0.6918489073968943, | |
| "learning_rate": 2.5755915538425315e-06, | |
| "loss": 0.1885, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10040745884180069, | |
| "step": 5160, | |
| "valid_targets_mean": 2994.4, | |
| "valid_targets_min": 1039 | |
| }, | |
| { | |
| "epoch": 5.978575564562826, | |
| "grad_norm": 0.6954095340384442, | |
| "learning_rate": 2.5473308291994657e-06, | |
| "loss": 0.1991, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12377315014600754, | |
| "step": 5165, | |
| "valid_targets_mean": 4625.4, | |
| "valid_targets_min": 2526 | |
| }, | |
| { | |
| "epoch": 5.984365952518819, | |
| "grad_norm": 0.6709528127757199, | |
| "learning_rate": 2.5192154577707983e-06, | |
| "loss": 0.194, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10651902109384537, | |
| "step": 5170, | |
| "valid_targets_mean": 3571.8, | |
| "valid_targets_min": 2409 | |
| }, | |
| { | |
| "epoch": 5.990156340474812, | |
| "grad_norm": 0.6603236277053569, | |
| "learning_rate": 2.4912456737132428e-06, | |
| "loss": 0.1943, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08352956175804138, | |
| "step": 5175, | |
| "valid_targets_mean": 3691.5, | |
| "valid_targets_min": 2918 | |
| }, | |
| { | |
| "epoch": 5.995946728430805, | |
| "grad_norm": 0.682482921886044, | |
| "learning_rate": 2.463421709971001e-06, | |
| "loss": 0.1997, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0977027416229248, | |
| "step": 5180, | |
| "valid_targets_mean": 4063.1, | |
| "valid_targets_min": 2179 | |
| }, | |
| { | |
| "epoch": 6.001158077591199, | |
| "grad_norm": 0.6295607542604371, | |
| "learning_rate": 2.4357437982738107e-06, | |
| "loss": 0.1826, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0902089774608612, | |
| "step": 5185, | |
| "valid_targets_mean": 8215.2, | |
| "valid_targets_min": 6163 | |
| }, | |
| { | |
| "epoch": 6.006948465547191, | |
| "grad_norm": 0.6689994894967519, | |
| "learning_rate": 2.408212169135031e-06, | |
| "loss": 0.1797, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09361856430768967, | |
| "step": 5190, | |
| "valid_targets_mean": 8594.4, | |
| "valid_targets_min": 6006 | |
| }, | |
| { | |
| "epoch": 6.012738853503185, | |
| "grad_norm": 0.645912983139216, | |
| "learning_rate": 2.3808270518497258e-06, | |
| "loss": 0.1689, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07413487136363983, | |
| "step": 5195, | |
| "valid_targets_mean": 3578.4, | |
| "valid_targets_min": 289 | |
| }, | |
| { | |
| "epoch": 6.018529241459178, | |
| "grad_norm": 0.5057537450377909, | |
| "learning_rate": 2.3535886744927373e-06, | |
| "loss": 0.17, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08606113493442535, | |
| "step": 5200, | |
| "valid_targets_mean": 7019.9, | |
| "valid_targets_min": 4985 | |
| }, | |
| { | |
| "epoch": 6.024319629415171, | |
| "grad_norm": 0.6413968314620463, | |
| "learning_rate": 2.3264972639168005e-06, | |
| "loss": 0.1632, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08514834195375443, | |
| "step": 5205, | |
| "valid_targets_mean": 6956.6, | |
| "valid_targets_min": 1457 | |
| }, | |
| { | |
| "epoch": 6.030110017371164, | |
| "grad_norm": 0.43875479622506836, | |
| "learning_rate": 2.2995530457506463e-06, | |
| "loss": 0.1672, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07854004204273224, | |
| "step": 5210, | |
| "valid_targets_mean": 7361.6, | |
| "valid_targets_min": 4961 | |
| }, | |
| { | |
| "epoch": 6.0359004053271565, | |
| "grad_norm": 0.5345119958764212, | |
| "learning_rate": 2.272756244397132e-06, | |
| "loss": 0.1822, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10500028729438782, | |
| "step": 5215, | |
| "valid_targets_mean": 7058.5, | |
| "valid_targets_min": 5985 | |
| }, | |
| { | |
| "epoch": 6.04169079328315, | |
| "grad_norm": 0.44174736329293024, | |
| "learning_rate": 2.2461070830313457e-06, | |
| "loss": 0.1748, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09389522671699524, | |
| "step": 5220, | |
| "valid_targets_mean": 7664.2, | |
| "valid_targets_min": 5169 | |
| }, | |
| { | |
| "epoch": 6.047481181239143, | |
| "grad_norm": 0.4695724589920154, | |
| "learning_rate": 2.219605783598795e-06, | |
| "loss": 0.1643, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08354824036359787, | |
| "step": 5225, | |
| "valid_targets_mean": 7111.8, | |
| "valid_targets_min": 5348 | |
| }, | |
| { | |
| "epoch": 6.053271569195136, | |
| "grad_norm": 0.5107050185858826, | |
| "learning_rate": 2.193252566813504e-06, | |
| "loss": 0.1524, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0697193592786789, | |
| "step": 5230, | |
| "valid_targets_mean": 4238.9, | |
| "valid_targets_min": 152 | |
| }, | |
| { | |
| "epoch": 6.059061957151129, | |
| "grad_norm": 0.4670961151455812, | |
| "learning_rate": 2.1670476521562202e-06, | |
| "loss": 0.1571, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08529293537139893, | |
| "step": 5235, | |
| "valid_targets_mean": 7199.4, | |
| "valid_targets_min": 4258 | |
| }, | |
| { | |
| "epoch": 6.064852345107123, | |
| "grad_norm": 0.5173124026480072, | |
| "learning_rate": 2.1409912578725624e-06, | |
| "loss": 0.1641, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0789559930562973, | |
| "step": 5240, | |
| "valid_targets_mean": 7243.4, | |
| "valid_targets_min": 5831 | |
| }, | |
| { | |
| "epoch": 6.070642733063115, | |
| "grad_norm": 0.4614128874803108, | |
| "learning_rate": 2.115083600971206e-06, | |
| "loss": 0.1618, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0823356881737709, | |
| "step": 5245, | |
| "valid_targets_mean": 6950.6, | |
| "valid_targets_min": 3862 | |
| }, | |
| { | |
| "epoch": 6.076433121019108, | |
| "grad_norm": 0.5532390387354605, | |
| "learning_rate": 2.089324897222087e-06, | |
| "loss": 0.1611, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08550730347633362, | |
| "step": 5250, | |
| "valid_targets_mean": 6991.9, | |
| "valid_targets_min": 5696 | |
| }, | |
| { | |
| "epoch": 6.082223508975101, | |
| "grad_norm": 0.44910028703040206, | |
| "learning_rate": 2.0637153611545835e-06, | |
| "loss": 0.1554, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0870165005326271, | |
| "step": 5255, | |
| "valid_targets_mean": 6963.5, | |
| "valid_targets_min": 4947 | |
| }, | |
| { | |
| "epoch": 6.088013896931094, | |
| "grad_norm": 0.4295536500825577, | |
| "learning_rate": 2.0382552060557525e-06, | |
| "loss": 0.1601, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07566852867603302, | |
| "step": 5260, | |
| "valid_targets_mean": 6436.9, | |
| "valid_targets_min": 4988 | |
| }, | |
| { | |
| "epoch": 6.093804284887088, | |
| "grad_norm": 0.43376542972026405, | |
| "learning_rate": 2.012944643968542e-06, | |
| "loss": 0.1621, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07929432392120361, | |
| "step": 5265, | |
| "valid_targets_mean": 7518.0, | |
| "valid_targets_min": 5476 | |
| }, | |
| { | |
| "epoch": 6.0995946728430805, | |
| "grad_norm": 0.39792800727485456, | |
| "learning_rate": 1.9877838856900177e-06, | |
| "loss": 0.1589, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07714049518108368, | |
| "step": 5270, | |
| "valid_targets_mean": 7422.6, | |
| "valid_targets_min": 5226 | |
| }, | |
| { | |
| "epoch": 6.105385060799073, | |
| "grad_norm": 0.4879186264265735, | |
| "learning_rate": 1.962773140769627e-06, | |
| "loss": 0.1681, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10403196513652802, | |
| "step": 5275, | |
| "valid_targets_mean": 7080.0, | |
| "valid_targets_min": 3753 | |
| }, | |
| { | |
| "epoch": 6.111175448755066, | |
| "grad_norm": 0.5775563776522831, | |
| "learning_rate": 1.9379126175074338e-06, | |
| "loss": 0.1818, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09144257009029388, | |
| "step": 5280, | |
| "valid_targets_mean": 6064.6, | |
| "valid_targets_min": 5161 | |
| }, | |
| { | |
| "epoch": 6.11696583671106, | |
| "grad_norm": 0.44043326178371406, | |
| "learning_rate": 1.9132025229524045e-06, | |
| "loss": 0.1301, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07593751698732376, | |
| "step": 5285, | |
| "valid_targets_mean": 7388.1, | |
| "valid_targets_min": 5795 | |
| }, | |
| { | |
| "epoch": 6.122756224667053, | |
| "grad_norm": 0.43469795273500295, | |
| "learning_rate": 1.8886430629006524e-06, | |
| "loss": 0.1553, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07400475442409515, | |
| "step": 5290, | |
| "valid_targets_mean": 7696.0, | |
| "valid_targets_min": 4658 | |
| }, | |
| { | |
| "epoch": 6.128546612623046, | |
| "grad_norm": 0.45002670649485194, | |
| "learning_rate": 1.8642344418937597e-06, | |
| "loss": 0.1578, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08555644005537033, | |
| "step": 5295, | |
| "valid_targets_mean": 7799.1, | |
| "valid_targets_min": 5508 | |
| }, | |
| { | |
| "epoch": 6.1343370005790385, | |
| "grad_norm": 0.4678842267897686, | |
| "learning_rate": 1.83997686321705e-06, | |
| "loss": 0.1665, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08118224889039993, | |
| "step": 5300, | |
| "valid_targets_mean": 7341.6, | |
| "valid_targets_min": 6454 | |
| }, | |
| { | |
| "epoch": 6.140127388535032, | |
| "grad_norm": 0.4525273642891633, | |
| "learning_rate": 1.8158705288979073e-06, | |
| "loss": 0.1604, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07110708206892014, | |
| "step": 5305, | |
| "valid_targets_mean": 6818.0, | |
| "valid_targets_min": 4392 | |
| }, | |
| { | |
| "epoch": 6.145917776491025, | |
| "grad_norm": 0.4298419040599829, | |
| "learning_rate": 1.7919156397040782e-06, | |
| "loss": 0.1592, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.06920880824327469, | |
| "step": 5310, | |
| "valid_targets_mean": 7857.6, | |
| "valid_targets_min": 5394 | |
| }, | |
| { | |
| "epoch": 6.151708164447018, | |
| "grad_norm": 0.39182340853570996, | |
| "learning_rate": 1.768112395142021e-06, | |
| "loss": 0.1414, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.06716249883174896, | |
| "step": 5315, | |
| "valid_targets_mean": 8314.1, | |
| "valid_targets_min": 4837 | |
| }, | |
| { | |
| "epoch": 6.157498552403011, | |
| "grad_norm": 0.3775500815963387, | |
| "learning_rate": 1.7444609934552326e-06, | |
| "loss": 0.1417, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.06070984899997711, | |
| "step": 5320, | |
| "valid_targets_mean": 7368.2, | |
| "valid_targets_min": 6027 | |
| }, | |
| { | |
| "epoch": 6.163288940359004, | |
| "grad_norm": 0.4483137843277887, | |
| "learning_rate": 1.7209616316225863e-06, | |
| "loss": 0.1491, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07993191480636597, | |
| "step": 5325, | |
| "valid_targets_mean": 6435.0, | |
| "valid_targets_min": 4238 | |
| }, | |
| { | |
| "epoch": 6.169079328314997, | |
| "grad_norm": 1.4199087279438916, | |
| "learning_rate": 1.6976145053567172e-06, | |
| "loss": 0.1438, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.04002588987350464, | |
| "step": 5330, | |
| "valid_targets_mean": 247.8, | |
| "valid_targets_min": 137 | |
| }, | |
| { | |
| "epoch": 6.17486971627099, | |
| "grad_norm": 0.45861727354707577, | |
| "learning_rate": 1.6744198091023745e-06, | |
| "loss": 0.1647, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07827004045248032, | |
| "step": 5335, | |
| "valid_targets_mean": 6807.0, | |
| "valid_targets_min": 5165 | |
| }, | |
| { | |
| "epoch": 6.180660104226983, | |
| "grad_norm": 0.5518915576183956, | |
| "learning_rate": 1.6513777360348005e-06, | |
| "loss": 0.1734, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09153981506824493, | |
| "step": 5340, | |
| "valid_targets_mean": 6989.9, | |
| "valid_targets_min": 5104 | |
| }, | |
| { | |
| "epoch": 6.186450492182976, | |
| "grad_norm": 0.48283612871721937, | |
| "learning_rate": 1.6284884780581345e-06, | |
| "loss": 0.1697, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09221628308296204, | |
| "step": 5345, | |
| "valid_targets_mean": 6692.2, | |
| "valid_targets_min": 5068 | |
| }, | |
| { | |
| "epoch": 6.19224088013897, | |
| "grad_norm": 0.47390055367830936, | |
| "learning_rate": 1.6057522258037982e-06, | |
| "loss": 0.1663, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09916602820158005, | |
| "step": 5350, | |
| "valid_targets_mean": 7419.6, | |
| "valid_targets_min": 5578 | |
| }, | |
| { | |
| "epoch": 6.1980312680949625, | |
| "grad_norm": 0.43306006927650187, | |
| "learning_rate": 1.5831691686289263e-06, | |
| "loss": 0.1598, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08104896545410156, | |
| "step": 5355, | |
| "valid_targets_mean": 7536.5, | |
| "valid_targets_min": 5075 | |
| }, | |
| { | |
| "epoch": 6.203821656050955, | |
| "grad_norm": 0.4064542024075362, | |
| "learning_rate": 1.5607394946147713e-06, | |
| "loss": 0.1667, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07590160518884659, | |
| "step": 5360, | |
| "valid_targets_mean": 7221.5, | |
| "valid_targets_min": 5841 | |
| }, | |
| { | |
| "epoch": 6.209612044006948, | |
| "grad_norm": 0.40126692667989083, | |
| "learning_rate": 1.5384633905651524e-06, | |
| "loss": 0.1609, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07777977734804153, | |
| "step": 5365, | |
| "valid_targets_mean": 7287.5, | |
| "valid_targets_min": 5151 | |
| }, | |
| { | |
| "epoch": 6.215402431962941, | |
| "grad_norm": 0.46137222872969186, | |
| "learning_rate": 1.5163410420048897e-06, | |
| "loss": 0.1702, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07954985648393631, | |
| "step": 5370, | |
| "valid_targets_mean": 6738.4, | |
| "valid_targets_min": 4042 | |
| }, | |
| { | |
| "epoch": 6.221192819918935, | |
| "grad_norm": 0.4409684532102382, | |
| "learning_rate": 1.4943726331782672e-06, | |
| "loss": 0.1644, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0905829519033432, | |
| "step": 5375, | |
| "valid_targets_mean": 6331.0, | |
| "valid_targets_min": 5496 | |
| }, | |
| { | |
| "epoch": 6.226983207874928, | |
| "grad_norm": 0.4553367903867736, | |
| "learning_rate": 1.4725583470474835e-06, | |
| "loss": 0.165, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0791117399930954, | |
| "step": 5380, | |
| "valid_targets_mean": 6271.1, | |
| "valid_targets_min": 4695 | |
| }, | |
| { | |
| "epoch": 6.2327735958309205, | |
| "grad_norm": 0.48629704290044623, | |
| "learning_rate": 1.4508983652911558e-06, | |
| "loss": 0.1623, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08450581133365631, | |
| "step": 5385, | |
| "valid_targets_mean": 5790.8, | |
| "valid_targets_min": 5021 | |
| }, | |
| { | |
| "epoch": 6.238563983786913, | |
| "grad_norm": 0.530857423794274, | |
| "learning_rate": 1.4293928683027681e-06, | |
| "loss": 0.1555, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0609971359372139, | |
| "step": 5390, | |
| "valid_targets_mean": 3386.8, | |
| "valid_targets_min": 2279 | |
| }, | |
| { | |
| "epoch": 6.244354371742907, | |
| "grad_norm": 0.43948413393234587, | |
| "learning_rate": 1.408042035189201e-06, | |
| "loss": 0.1352, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07962469756603241, | |
| "step": 5395, | |
| "valid_targets_mean": 6240.5, | |
| "valid_targets_min": 5093 | |
| }, | |
| { | |
| "epoch": 6.2501447596989, | |
| "grad_norm": 0.4649474702410835, | |
| "learning_rate": 1.3868460437692298e-06, | |
| "loss": 0.1599, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08120064437389374, | |
| "step": 5400, | |
| "valid_targets_mean": 6329.5, | |
| "valid_targets_min": 4589 | |
| }, | |
| { | |
| "epoch": 6.255935147654893, | |
| "grad_norm": 0.44995107882935337, | |
| "learning_rate": 1.3658050705720417e-06, | |
| "loss": 0.1617, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09052136540412903, | |
| "step": 5405, | |
| "valid_targets_mean": 7589.2, | |
| "valid_targets_min": 4963 | |
| }, | |
| { | |
| "epoch": 6.261725535610886, | |
| "grad_norm": 0.441803859460047, | |
| "learning_rate": 1.3449192908357622e-06, | |
| "loss": 0.1594, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.06894394010305405, | |
| "step": 5410, | |
| "valid_targets_mean": 5577.9, | |
| "valid_targets_min": 4458 | |
| }, | |
| { | |
| "epoch": 6.267515923566879, | |
| "grad_norm": 0.4786237029391992, | |
| "learning_rate": 1.3241888785060076e-06, | |
| "loss": 0.1595, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07872607558965683, | |
| "step": 5415, | |
| "valid_targets_mean": 5432.5, | |
| "valid_targets_min": 4433 | |
| }, | |
| { | |
| "epoch": 6.273306311522872, | |
| "grad_norm": 0.46161869669093514, | |
| "learning_rate": 1.3036140062344215e-06, | |
| "loss": 0.1632, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08250191807746887, | |
| "step": 5420, | |
| "valid_targets_mean": 6361.8, | |
| "valid_targets_min": 5201 | |
| }, | |
| { | |
| "epoch": 6.279096699478865, | |
| "grad_norm": 0.48405430754178314, | |
| "learning_rate": 1.2831948453772514e-06, | |
| "loss": 0.1627, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07658439129590988, | |
| "step": 5425, | |
| "valid_targets_mean": 5824.4, | |
| "valid_targets_min": 5129 | |
| }, | |
| { | |
| "epoch": 6.284887087434858, | |
| "grad_norm": 0.8564807909910727, | |
| "learning_rate": 1.2629315659939056e-06, | |
| "loss": 0.1242, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07406964898109436, | |
| "step": 5430, | |
| "valid_targets_mean": 1637.6, | |
| "valid_targets_min": 137 | |
| }, | |
| { | |
| "epoch": 6.290677475390851, | |
| "grad_norm": 0.4606197263140385, | |
| "learning_rate": 1.2428243368455472e-06, | |
| "loss": 0.1563, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07980707287788391, | |
| "step": 5435, | |
| "valid_targets_mean": 6474.6, | |
| "valid_targets_min": 4751 | |
| }, | |
| { | |
| "epoch": 6.2964678633468445, | |
| "grad_norm": 0.4596236677668687, | |
| "learning_rate": 1.2228733253936942e-06, | |
| "loss": 0.1593, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08767077326774597, | |
| "step": 5440, | |
| "valid_targets_mean": 7186.2, | |
| "valid_targets_min": 5358 | |
| }, | |
| { | |
| "epoch": 6.302258251302837, | |
| "grad_norm": 0.4448274823182163, | |
| "learning_rate": 1.2030786977988095e-06, | |
| "loss": 0.1625, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08086740225553513, | |
| "step": 5445, | |
| "valid_targets_mean": 7126.8, | |
| "valid_targets_min": 4078 | |
| }, | |
| { | |
| "epoch": 6.30804863925883, | |
| "grad_norm": 0.4338103596308614, | |
| "learning_rate": 1.1834406189189297e-06, | |
| "loss": 0.1541, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07633671909570694, | |
| "step": 5450, | |
| "valid_targets_mean": 7022.1, | |
| "valid_targets_min": 4940 | |
| }, | |
| { | |
| "epoch": 6.313839027214823, | |
| "grad_norm": 0.47759713239926815, | |
| "learning_rate": 1.1639592523082865e-06, | |
| "loss": 0.1518, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08213866502046585, | |
| "step": 5455, | |
| "valid_targets_mean": 5900.6, | |
| "valid_targets_min": 4779 | |
| }, | |
| { | |
| "epoch": 6.319629415170817, | |
| "grad_norm": 0.45111566647701096, | |
| "learning_rate": 1.1446347602159459e-06, | |
| "loss": 0.1578, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08208338916301727, | |
| "step": 5460, | |
| "valid_targets_mean": 6290.1, | |
| "valid_targets_min": 4620 | |
| }, | |
| { | |
| "epoch": 6.32541980312681, | |
| "grad_norm": 0.6232495525960332, | |
| "learning_rate": 1.1254673035844578e-06, | |
| "loss": 0.195, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10899636149406433, | |
| "step": 5465, | |
| "valid_targets_mean": 5614.5, | |
| "valid_targets_min": 3446 | |
| }, | |
| { | |
| "epoch": 6.3312101910828025, | |
| "grad_norm": 0.6948828112562729, | |
| "learning_rate": 1.1064570420485076e-06, | |
| "loss": 0.2084, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11155211180448532, | |
| "step": 5470, | |
| "valid_targets_mean": 4630.0, | |
| "valid_targets_min": 1005 | |
| }, | |
| { | |
| "epoch": 6.337000579038795, | |
| "grad_norm": 0.5438326454147281, | |
| "learning_rate": 1.0876041339335974e-06, | |
| "loss": 0.2072, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07827803492546082, | |
| "step": 5475, | |
| "valid_targets_mean": 4669.6, | |
| "valid_targets_min": 2054 | |
| }, | |
| { | |
| "epoch": 6.342790966994789, | |
| "grad_norm": 0.6097818884247744, | |
| "learning_rate": 1.068908736254728e-06, | |
| "loss": 0.2023, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08751150965690613, | |
| "step": 5480, | |
| "valid_targets_mean": 3745.0, | |
| "valid_targets_min": 1561 | |
| }, | |
| { | |
| "epoch": 6.348581354950782, | |
| "grad_norm": 0.601122129840999, | |
| "learning_rate": 1.0503710047150783e-06, | |
| "loss": 0.2032, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10058531910181046, | |
| "step": 5485, | |
| "valid_targets_mean": 4284.0, | |
| "valid_targets_min": 1184 | |
| }, | |
| { | |
| "epoch": 6.354371742906775, | |
| "grad_norm": 0.5355235715879467, | |
| "learning_rate": 1.031991093704725e-06, | |
| "loss": 0.2107, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1052846908569336, | |
| "step": 5490, | |
| "valid_targets_mean": 6332.8, | |
| "valid_targets_min": 1651 | |
| }, | |
| { | |
| "epoch": 6.360162130862768, | |
| "grad_norm": 0.5356930805063133, | |
| "learning_rate": 1.0137691562993447e-06, | |
| "loss": 0.203, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10232044756412506, | |
| "step": 5495, | |
| "valid_targets_mean": 6446.5, | |
| "valid_targets_min": 2231 | |
| }, | |
| { | |
| "epoch": 6.3659525188187605, | |
| "grad_norm": 0.6234673466277105, | |
| "learning_rate": 9.957053442589436e-07, | |
| "loss": 0.1999, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07552198320627213, | |
| "step": 5500, | |
| "valid_targets_mean": 3367.2, | |
| "valid_targets_min": 467 | |
| }, | |
| { | |
| "epoch": 6.371742906774754, | |
| "grad_norm": 0.6370748608739049, | |
| "learning_rate": 9.777998080265893e-07, | |
| "loss": 0.207, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13317011296749115, | |
| "step": 5505, | |
| "valid_targets_mean": 7194.1, | |
| "valid_targets_min": 1748 | |
| }, | |
| { | |
| "epoch": 6.377533294730747, | |
| "grad_norm": 0.6441957718117712, | |
| "learning_rate": 9.600526967271696e-07, | |
| "loss": 0.2107, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10556194186210632, | |
| "step": 5510, | |
| "valid_targets_mean": 4078.5, | |
| "valid_targets_min": 1697 | |
| }, | |
| { | |
| "epoch": 6.38332368268674, | |
| "grad_norm": 0.6850420900180021, | |
| "learning_rate": 9.42464158166132e-07, | |
| "loss": 0.204, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.06766747683286667, | |
| "step": 5515, | |
| "valid_targets_mean": 2089.5, | |
| "valid_targets_min": 1223 | |
| }, | |
| { | |
| "epoch": 6.389114070642733, | |
| "grad_norm": 0.5835185801774779, | |
| "learning_rate": 9.250343388282701e-07, | |
| "loss": 0.205, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09306862950325012, | |
| "step": 5520, | |
| "valid_targets_mean": 5404.1, | |
| "valid_targets_min": 1829 | |
| }, | |
| { | |
| "epoch": 6.3949044585987265, | |
| "grad_norm": 0.5675457701788594, | |
| "learning_rate": 9.077633838764987e-07, | |
| "loss": 0.2072, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09642751514911652, | |
| "step": 5525, | |
| "valid_targets_mean": 4583.1, | |
| "valid_targets_min": 1596 | |
| }, | |
| { | |
| "epoch": 6.400694846554719, | |
| "grad_norm": 0.6202832830411663, | |
| "learning_rate": 8.906514371506358e-07, | |
| "loss": 0.1995, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10527827590703964, | |
| "step": 5530, | |
| "valid_targets_mean": 4666.8, | |
| "valid_targets_min": 2041 | |
| }, | |
| { | |
| "epoch": 6.406485234510712, | |
| "grad_norm": 0.6361782574350211, | |
| "learning_rate": 8.736986411662185e-07, | |
| "loss": 0.1985, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1075083315372467, | |
| "step": 5535, | |
| "valid_targets_mean": 4272.1, | |
| "valid_targets_min": 1538 | |
| }, | |
| { | |
| "epoch": 6.412275622466705, | |
| "grad_norm": 0.6947641383759208, | |
| "learning_rate": 8.569051371133019e-07, | |
| "loss": 0.1946, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08818969130516052, | |
| "step": 5540, | |
| "valid_targets_mean": 2928.0, | |
| "valid_targets_min": 1739 | |
| }, | |
| { | |
| "epoch": 6.418066010422699, | |
| "grad_norm": 0.6786378448146124, | |
| "learning_rate": 8.402710648552958e-07, | |
| "loss": 0.1962, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09193718433380127, | |
| "step": 5545, | |
| "valid_targets_mean": 3254.5, | |
| "valid_targets_min": 935 | |
| }, | |
| { | |
| "epoch": 6.423856398378692, | |
| "grad_norm": 0.67462549653229, | |
| "learning_rate": 8.237965629277967e-07, | |
| "loss": 0.2008, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1017710268497467, | |
| "step": 5550, | |
| "valid_targets_mean": 3535.2, | |
| "valid_targets_min": 2704 | |
| }, | |
| { | |
| "epoch": 6.4296467863346844, | |
| "grad_norm": 0.7166248619877422, | |
| "learning_rate": 8.074817685374215e-07, | |
| "loss": 0.2002, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09396373480558395, | |
| "step": 5555, | |
| "valid_targets_mean": 2987.6, | |
| "valid_targets_min": 1173 | |
| }, | |
| { | |
| "epoch": 6.435437174290677, | |
| "grad_norm": 0.6380381399423711, | |
| "learning_rate": 7.913268175606869e-07, | |
| "loss": 0.2041, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0939275324344635, | |
| "step": 5560, | |
| "valid_targets_mean": 3700.6, | |
| "valid_targets_min": 1670 | |
| }, | |
| { | |
| "epoch": 6.44122756224667, | |
| "grad_norm": 0.6878641353617551, | |
| "learning_rate": 7.753318445428571e-07, | |
| "loss": 0.1994, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08689321577548981, | |
| "step": 5565, | |
| "valid_targets_mean": 3481.6, | |
| "valid_targets_min": 1329 | |
| }, | |
| { | |
| "epoch": 6.447017950202664, | |
| "grad_norm": 0.675521361398003, | |
| "learning_rate": 7.594969826968435e-07, | |
| "loss": 0.1925, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10225418955087662, | |
| "step": 5570, | |
| "valid_targets_mean": 3589.8, | |
| "valid_targets_min": 2279 | |
| }, | |
| { | |
| "epoch": 6.452808338158657, | |
| "grad_norm": 0.5928946044089235, | |
| "learning_rate": 7.43822363902067e-07, | |
| "loss": 0.1813, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09262774884700775, | |
| "step": 5575, | |
| "valid_targets_mean": 5060.4, | |
| "valid_targets_min": 2222 | |
| }, | |
| { | |
| "epoch": 6.45859872611465, | |
| "grad_norm": 0.6509966309795416, | |
| "learning_rate": 7.283081187033914e-07, | |
| "loss": 0.1921, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0970219075679779, | |
| "step": 5580, | |
| "valid_targets_mean": 3898.6, | |
| "valid_targets_min": 1849 | |
| }, | |
| { | |
| "epoch": 6.464389114070642, | |
| "grad_norm": 0.6272129163596066, | |
| "learning_rate": 7.129543763100155e-07, | |
| "loss": 0.1974, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08521132171154022, | |
| "step": 5585, | |
| "valid_targets_mean": 3698.9, | |
| "valid_targets_min": 1951 | |
| }, | |
| { | |
| "epoch": 6.470179502026636, | |
| "grad_norm": 0.6416525439771972, | |
| "learning_rate": 6.977612645944031e-07, | |
| "loss": 0.2003, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10487203299999237, | |
| "step": 5590, | |
| "valid_targets_mean": 4506.8, | |
| "valid_targets_min": 1765 | |
| }, | |
| { | |
| "epoch": 6.475969889982629, | |
| "grad_norm": 0.6986484122660109, | |
| "learning_rate": 6.827289100912171e-07, | |
| "loss": 0.1873, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09219237416982651, | |
| "step": 5595, | |
| "valid_targets_mean": 2991.6, | |
| "valid_targets_min": 2087 | |
| }, | |
| { | |
| "epoch": 6.481760277938622, | |
| "grad_norm": 0.6558910975438434, | |
| "learning_rate": 6.678574379962666e-07, | |
| "loss": 0.1887, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08751760423183441, | |
| "step": 5600, | |
| "valid_targets_mean": 3994.1, | |
| "valid_targets_min": 1603 | |
| }, | |
| { | |
| "epoch": 6.487550665894615, | |
| "grad_norm": 0.646081026641617, | |
| "learning_rate": 6.531469721654682e-07, | |
| "loss": 0.1928, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10096277296543121, | |
| "step": 5605, | |
| "valid_targets_mean": 4391.9, | |
| "valid_targets_min": 2092 | |
| }, | |
| { | |
| "epoch": 6.4933410538506084, | |
| "grad_norm": 0.6444455652055436, | |
| "learning_rate": 6.385976351138024e-07, | |
| "loss": 0.1878, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09697127342224121, | |
| "step": 5610, | |
| "valid_targets_mean": 3808.9, | |
| "valid_targets_min": 1410 | |
| }, | |
| { | |
| "epoch": 6.499131441806601, | |
| "grad_norm": 0.6125684704713852, | |
| "learning_rate": 6.242095480143051e-07, | |
| "loss": 0.1827, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09984603524208069, | |
| "step": 5615, | |
| "valid_targets_mean": 5078.5, | |
| "valid_targets_min": 2110 | |
| }, | |
| { | |
| "epoch": 6.504921829762594, | |
| "grad_norm": 0.6864467278013624, | |
| "learning_rate": 6.099828306970579e-07, | |
| "loss": 0.1935, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09133944660425186, | |
| "step": 5620, | |
| "valid_targets_mean": 3447.4, | |
| "valid_targets_min": 2420 | |
| }, | |
| { | |
| "epoch": 6.510712217718587, | |
| "grad_norm": 0.6369169118299097, | |
| "learning_rate": 5.959176016481816e-07, | |
| "loss": 0.1884, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08592967689037323, | |
| "step": 5625, | |
| "valid_targets_mean": 4371.1, | |
| "valid_targets_min": 1876 | |
| }, | |
| { | |
| "epoch": 6.51650260567458, | |
| "grad_norm": 0.6711562549633678, | |
| "learning_rate": 5.82013978008853e-07, | |
| "loss": 0.188, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10728715360164642, | |
| "step": 5630, | |
| "valid_targets_mean": 3858.2, | |
| "valid_targets_min": 2605 | |
| }, | |
| { | |
| "epoch": 6.522292993630574, | |
| "grad_norm": 0.7147202515551577, | |
| "learning_rate": 5.682720755743342e-07, | |
| "loss": 0.1893, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08998127281665802, | |
| "step": 5635, | |
| "valid_targets_mean": 3383.9, | |
| "valid_targets_min": 1177 | |
| }, | |
| { | |
| "epoch": 6.528083381586566, | |
| "grad_norm": 0.7321320344926183, | |
| "learning_rate": 5.546920087930097e-07, | |
| "loss": 0.1913, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09720846265554428, | |
| "step": 5640, | |
| "valid_targets_mean": 3455.5, | |
| "valid_targets_min": 1714 | |
| }, | |
| { | |
| "epoch": 6.533873769542559, | |
| "grad_norm": 0.6270437609775642, | |
| "learning_rate": 5.412738907654147e-07, | |
| "loss": 0.1919, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10687099397182465, | |
| "step": 5645, | |
| "valid_targets_mean": 5547.5, | |
| "valid_targets_min": 2033 | |
| }, | |
| { | |
| "epoch": 6.539664157498552, | |
| "grad_norm": 1.3739530314415442, | |
| "learning_rate": 5.280178332433173e-07, | |
| "loss": 0.1862, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0967320129275322, | |
| "step": 5650, | |
| "valid_targets_mean": 3485.4, | |
| "valid_targets_min": 1924 | |
| }, | |
| { | |
| "epoch": 6.545454545454545, | |
| "grad_norm": 0.7015965787431185, | |
| "learning_rate": 5.149239466287736e-07, | |
| "loss": 0.1949, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08975489437580109, | |
| "step": 5655, | |
| "valid_targets_mean": 3080.2, | |
| "valid_targets_min": 2098 | |
| }, | |
| { | |
| "epoch": 6.551244933410539, | |
| "grad_norm": 0.7320039568179678, | |
| "learning_rate": 5.01992339973214e-07, | |
| "loss": 0.1909, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10681869834661484, | |
| "step": 5660, | |
| "valid_targets_mean": 3393.4, | |
| "valid_targets_min": 1363 | |
| }, | |
| { | |
| "epoch": 6.5570353213665316, | |
| "grad_norm": 0.6658995441329661, | |
| "learning_rate": 4.892231209765341e-07, | |
| "loss": 0.1889, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10230939090251923, | |
| "step": 5665, | |
| "valid_targets_mean": 4115.4, | |
| "valid_targets_min": 1363 | |
| }, | |
| { | |
| "epoch": 6.562825709322524, | |
| "grad_norm": 0.6514909570196014, | |
| "learning_rate": 4.7661639598618467e-07, | |
| "loss": 0.1908, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1146596372127533, | |
| "step": 5670, | |
| "valid_targets_mean": 4313.8, | |
| "valid_targets_min": 2620 | |
| }, | |
| { | |
| "epoch": 6.568616097278518, | |
| "grad_norm": 0.7078908593122644, | |
| "learning_rate": 4.641722699963147e-07, | |
| "loss": 0.1927, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0996047705411911, | |
| "step": 5675, | |
| "valid_targets_mean": 3425.1, | |
| "valid_targets_min": 2316 | |
| }, | |
| { | |
| "epoch": 6.574406485234511, | |
| "grad_norm": 1.0795733147780582, | |
| "learning_rate": 4.5189084664686525e-07, | |
| "loss": 0.189, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09666218608617783, | |
| "step": 5680, | |
| "valid_targets_mean": 3744.0, | |
| "valid_targets_min": 2022 | |
| }, | |
| { | |
| "epoch": 6.580196873190504, | |
| "grad_norm": 0.6618911031171433, | |
| "learning_rate": 4.3977222822273013e-07, | |
| "loss": 0.1816, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0638245940208435, | |
| "step": 5685, | |
| "valid_targets_mean": 2445.6, | |
| "valid_targets_min": 1294 | |
| }, | |
| { | |
| "epoch": 6.585987261146497, | |
| "grad_norm": 0.7074342788867226, | |
| "learning_rate": 4.278165156528924e-07, | |
| "loss": 0.1924, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09848304837942123, | |
| "step": 5690, | |
| "valid_targets_mean": 3556.1, | |
| "valid_targets_min": 1700 | |
| }, | |
| { | |
| "epoch": 6.5917776491024895, | |
| "grad_norm": 0.6959478862926624, | |
| "learning_rate": 4.1602380850958247e-07, | |
| "loss": 0.1902, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08021435141563416, | |
| "step": 5695, | |
| "valid_targets_mean": 2714.0, | |
| "valid_targets_min": 1586 | |
| }, | |
| { | |
| "epoch": 6.597568037058483, | |
| "grad_norm": 0.734440751148546, | |
| "learning_rate": 4.043942050074612e-07, | |
| "loss": 0.1853, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09646306186914444, | |
| "step": 5700, | |
| "valid_targets_mean": 3266.4, | |
| "valid_targets_min": 1358 | |
| }, | |
| { | |
| "epoch": 6.603358425014476, | |
| "grad_norm": 0.7175842729357217, | |
| "learning_rate": 3.929278020027849e-07, | |
| "loss": 0.189, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0929139107465744, | |
| "step": 5705, | |
| "valid_targets_mean": 3782.9, | |
| "valid_targets_min": 2290 | |
| }, | |
| { | |
| "epoch": 6.609148812970469, | |
| "grad_norm": 0.7143653808913596, | |
| "learning_rate": 3.816246949926106e-07, | |
| "loss": 0.1964, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1292996108531952, | |
| "step": 5710, | |
| "valid_targets_mean": 4904.8, | |
| "valid_targets_min": 2189 | |
| }, | |
| { | |
| "epoch": 6.614939200926462, | |
| "grad_norm": 0.7001644859705362, | |
| "learning_rate": 3.704849781139941e-07, | |
| "loss": 0.1922, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11987542361021042, | |
| "step": 5715, | |
| "valid_targets_mean": 3959.5, | |
| "valid_targets_min": 2121 | |
| }, | |
| { | |
| "epoch": 6.620729588882455, | |
| "grad_norm": 0.8700422206648432, | |
| "learning_rate": 3.595087441432132e-07, | |
| "loss": 0.1925, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09515506774187088, | |
| "step": 5720, | |
| "valid_targets_mean": 3689.4, | |
| "valid_targets_min": 2011 | |
| }, | |
| { | |
| "epoch": 6.626519976838448, | |
| "grad_norm": 0.6443645086879763, | |
| "learning_rate": 3.4869608449499047e-07, | |
| "loss": 0.1846, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10317341238260269, | |
| "step": 5725, | |
| "valid_targets_mean": 5082.1, | |
| "valid_targets_min": 2674 | |
| }, | |
| { | |
| "epoch": 6.632310364794441, | |
| "grad_norm": 0.6934878123779786, | |
| "learning_rate": 3.380470892217314e-07, | |
| "loss": 0.1794, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10448549687862396, | |
| "step": 5730, | |
| "valid_targets_mean": 3660.9, | |
| "valid_targets_min": 2266 | |
| }, | |
| { | |
| "epoch": 6.638100752750434, | |
| "grad_norm": 0.6101536669987144, | |
| "learning_rate": 3.275618470127739e-07, | |
| "loss": 0.1851, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08371835947036743, | |
| "step": 5735, | |
| "valid_targets_mean": 4285.5, | |
| "valid_targets_min": 904 | |
| }, | |
| { | |
| "epoch": 6.643891140706427, | |
| "grad_norm": 0.6765235879602266, | |
| "learning_rate": 3.172404451936517e-07, | |
| "loss": 0.1893, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11076696217060089, | |
| "step": 5740, | |
| "valid_targets_mean": 3999.4, | |
| "valid_targets_min": 1194 | |
| }, | |
| { | |
| "epoch": 6.649681528662421, | |
| "grad_norm": 0.9097217500520384, | |
| "learning_rate": 3.0708296972536746e-07, | |
| "loss": 0.188, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08506985753774643, | |
| "step": 5745, | |
| "valid_targets_mean": 4011.8, | |
| "valid_targets_min": 1658 | |
| }, | |
| { | |
| "epoch": 6.6554719166184135, | |
| "grad_norm": 0.7605155604289844, | |
| "learning_rate": 2.970895052036782e-07, | |
| "loss": 0.1887, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0780138224363327, | |
| "step": 5750, | |
| "valid_targets_mean": 2728.8, | |
| "valid_targets_min": 1423 | |
| }, | |
| { | |
| "epoch": 6.661262304574406, | |
| "grad_norm": 0.6563600298248051, | |
| "learning_rate": 2.8726013485838033e-07, | |
| "loss": 0.1914, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10089323669672012, | |
| "step": 5755, | |
| "valid_targets_mean": 4504.4, | |
| "valid_targets_min": 2320 | |
| }, | |
| { | |
| "epoch": 6.667052692530399, | |
| "grad_norm": 0.6679398668890479, | |
| "learning_rate": 2.7759494055263237e-07, | |
| "loss": 0.1919, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07146115601062775, | |
| "step": 5760, | |
| "valid_targets_mean": 3272.5, | |
| "valid_targets_min": 1811 | |
| }, | |
| { | |
| "epoch": 6.672843080486393, | |
| "grad_norm": 0.6216942456939679, | |
| "learning_rate": 2.6809400278225983e-07, | |
| "loss": 0.1985, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09198446571826935, | |
| "step": 5765, | |
| "valid_targets_mean": 4425.9, | |
| "valid_targets_min": 3026 | |
| }, | |
| { | |
| "epoch": 6.678633468442386, | |
| "grad_norm": 0.677098777582908, | |
| "learning_rate": 2.5875740067509814e-07, | |
| "loss": 0.1875, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08467467874288559, | |
| "step": 5770, | |
| "valid_targets_mean": 2719.1, | |
| "valid_targets_min": 644 | |
| }, | |
| { | |
| "epoch": 6.684423856398379, | |
| "grad_norm": 0.7057087481731708, | |
| "learning_rate": 2.495852119903153e-07, | |
| "loss": 0.1837, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08754457533359528, | |
| "step": 5775, | |
| "valid_targets_mean": 3369.0, | |
| "valid_targets_min": 1397 | |
| }, | |
| { | |
| "epoch": 6.6902142443543715, | |
| "grad_norm": 0.6663376534097115, | |
| "learning_rate": 2.405775131177857e-07, | |
| "loss": 0.1761, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09364516288042068, | |
| "step": 5780, | |
| "valid_targets_mean": 4122.6, | |
| "valid_targets_min": 1415 | |
| }, | |
| { | |
| "epoch": 6.696004632310364, | |
| "grad_norm": 0.7057387566685178, | |
| "learning_rate": 2.317343790774329e-07, | |
| "loss": 0.1777, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09693238139152527, | |
| "step": 5785, | |
| "valid_targets_mean": 4136.6, | |
| "valid_targets_min": 621 | |
| }, | |
| { | |
| "epoch": 6.701795020266358, | |
| "grad_norm": 0.6662639138513196, | |
| "learning_rate": 2.2305588351862362e-07, | |
| "loss": 0.1807, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07178279757499695, | |
| "step": 5790, | |
| "valid_targets_mean": 3065.8, | |
| "valid_targets_min": 1290 | |
| }, | |
| { | |
| "epoch": 6.707585408222351, | |
| "grad_norm": 0.812162044772317, | |
| "learning_rate": 2.145420987195368e-07, | |
| "loss": 0.1693, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07713647931814194, | |
| "step": 5795, | |
| "valid_targets_mean": 2426.9, | |
| "valid_targets_min": 857 | |
| }, | |
| { | |
| "epoch": 6.713375796178344, | |
| "grad_norm": 0.8006317550584056, | |
| "learning_rate": 2.061930955865754e-07, | |
| "loss": 0.1737, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09362758696079254, | |
| "step": 5800, | |
| "valid_targets_mean": 2949.1, | |
| "valid_targets_min": 720 | |
| }, | |
| { | |
| "epoch": 6.719166184134337, | |
| "grad_norm": 0.7153682567907866, | |
| "learning_rate": 1.9800894365377355e-07, | |
| "loss": 0.1854, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10825473070144653, | |
| "step": 5805, | |
| "valid_targets_mean": 3579.1, | |
| "valid_targets_min": 1801 | |
| }, | |
| { | |
| "epoch": 6.72495657209033, | |
| "grad_norm": 0.6870837669469271, | |
| "learning_rate": 1.89989711082208e-07, | |
| "loss": 0.2007, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09038402140140533, | |
| "step": 5810, | |
| "valid_targets_mean": 3532.6, | |
| "valid_targets_min": 1510 | |
| }, | |
| { | |
| "epoch": 6.730746960046323, | |
| "grad_norm": 0.7410978699692073, | |
| "learning_rate": 1.8213546465944755e-07, | |
| "loss": 0.1764, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08246541023254395, | |
| "step": 5815, | |
| "valid_targets_mean": 2807.9, | |
| "valid_targets_min": 987 | |
| }, | |
| { | |
| "epoch": 6.736537348002316, | |
| "grad_norm": 0.6712440479339922, | |
| "learning_rate": 1.744462697989735e-07, | |
| "loss": 0.1924, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08393748104572296, | |
| "step": 5820, | |
| "valid_targets_mean": 3937.4, | |
| "valid_targets_min": 996 | |
| }, | |
| { | |
| "epoch": 6.742327735958309, | |
| "grad_norm": 0.6680483028378558, | |
| "learning_rate": 1.6692219053965786e-07, | |
| "loss": 0.1818, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07929886877536774, | |
| "step": 5825, | |
| "valid_targets_mean": 3954.5, | |
| "valid_targets_min": 2001 | |
| }, | |
| { | |
| "epoch": 6.748118123914303, | |
| "grad_norm": 0.5841660991413182, | |
| "learning_rate": 1.5956328954521928e-07, | |
| "loss": 0.1877, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07254894077777863, | |
| "step": 5830, | |
| "valid_targets_mean": 5059.6, | |
| "valid_targets_min": 1767 | |
| }, | |
| { | |
| "epoch": 6.7539085118702955, | |
| "grad_norm": 0.5919303114874458, | |
| "learning_rate": 1.5236962810369238e-07, | |
| "loss": 0.1856, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07777701318264008, | |
| "step": 5835, | |
| "valid_targets_mean": 4714.0, | |
| "valid_targets_min": 1511 | |
| }, | |
| { | |
| "epoch": 6.759698899826288, | |
| "grad_norm": 0.7289685243699716, | |
| "learning_rate": 1.4534126612693934e-07, | |
| "loss": 0.181, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08982829749584198, | |
| "step": 5840, | |
| "valid_targets_mean": 2861.0, | |
| "valid_targets_min": 974 | |
| }, | |
| { | |
| "epoch": 6.765489287782281, | |
| "grad_norm": 0.6489662405100123, | |
| "learning_rate": 1.3847826215013016e-07, | |
| "loss": 0.1846, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08057229220867157, | |
| "step": 5845, | |
| "valid_targets_mean": 3792.6, | |
| "valid_targets_min": 2339 | |
| }, | |
| { | |
| "epoch": 6.771279675738274, | |
| "grad_norm": 0.6709627866530247, | |
| "learning_rate": 1.3178067333126766e-07, | |
| "loss": 0.1799, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0797836035490036, | |
| "step": 5850, | |
| "valid_targets_mean": 3333.1, | |
| "valid_targets_min": 1249 | |
| }, | |
| { | |
| "epoch": 6.777070063694268, | |
| "grad_norm": 0.67605341835653, | |
| "learning_rate": 1.2524855545070548e-07, | |
| "loss": 0.1877, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09331607818603516, | |
| "step": 5855, | |
| "valid_targets_mean": 3493.1, | |
| "valid_targets_min": 1223 | |
| }, | |
| { | |
| "epoch": 6.782860451650261, | |
| "grad_norm": 0.755248924578272, | |
| "learning_rate": 1.188819629106841e-07, | |
| "loss": 0.1967, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09816304594278336, | |
| "step": 5860, | |
| "valid_targets_mean": 3285.8, | |
| "valid_targets_min": 1318 | |
| }, | |
| { | |
| "epoch": 6.7886508396062535, | |
| "grad_norm": 0.7842651252551903, | |
| "learning_rate": 1.1268094873488233e-07, | |
| "loss": 0.1898, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08679020404815674, | |
| "step": 5865, | |
| "valid_targets_mean": 2070.5, | |
| "valid_targets_min": 1002 | |
| }, | |
| { | |
| "epoch": 6.794441227562246, | |
| "grad_norm": 0.6986357267206671, | |
| "learning_rate": 1.0664556456797315e-07, | |
| "loss": 0.2041, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08979077637195587, | |
| "step": 5870, | |
| "valid_targets_mean": 3328.1, | |
| "valid_targets_min": 845 | |
| }, | |
| { | |
| "epoch": 6.80023161551824, | |
| "grad_norm": 0.65199188833835, | |
| "learning_rate": 1.0077586067518186e-07, | |
| "loss": 0.1818, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1040390133857727, | |
| "step": 5875, | |
| "valid_targets_mean": 4657.4, | |
| "valid_targets_min": 2747 | |
| }, | |
| { | |
| "epoch": 6.806022003474233, | |
| "grad_norm": 0.7163232433100037, | |
| "learning_rate": 9.507188594189309e-08, | |
| "loss": 0.2036, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1029033362865448, | |
| "step": 5880, | |
| "valid_targets_mean": 4166.4, | |
| "valid_targets_min": 1241 | |
| }, | |
| { | |
| "epoch": 6.811812391430226, | |
| "grad_norm": 0.7444738988192119, | |
| "learning_rate": 8.953368787322226e-08, | |
| "loss": 0.1959, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10908888280391693, | |
| "step": 5885, | |
| "valid_targets_mean": 3567.1, | |
| "valid_targets_min": 992 | |
| }, | |
| { | |
| "epoch": 6.817602779386219, | |
| "grad_norm": 0.6625460085007966, | |
| "learning_rate": 8.416131259362025e-08, | |
| "loss": 0.1929, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09041760861873627, | |
| "step": 5890, | |
| "valid_targets_mean": 3343.5, | |
| "valid_targets_min": 1211 | |
| }, | |
| { | |
| "epoch": 6.823393167342212, | |
| "grad_norm": 0.7809754980914095, | |
| "learning_rate": 7.895480484650941e-08, | |
| "loss": 0.1929, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.103350929915905, | |
| "step": 5895, | |
| "valid_targets_mean": 3081.6, | |
| "valid_targets_min": 1889 | |
| }, | |
| { | |
| "epoch": 6.829183555298205, | |
| "grad_norm": 0.7507747805313876, | |
| "learning_rate": 7.391420799388372e-08, | |
| "loss": 0.1912, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08229044079780579, | |
| "step": 5900, | |
| "valid_targets_mean": 3233.9, | |
| "valid_targets_min": 1624 | |
| }, | |
| { | |
| "epoch": 6.834973943254198, | |
| "grad_norm": 0.6766796999819782, | |
| "learning_rate": 6.903956401597134e-08, | |
| "loss": 0.1916, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10043347626924515, | |
| "step": 5905, | |
| "valid_targets_mean": 3817.4, | |
| "valid_targets_min": 2763 | |
| }, | |
| { | |
| "epoch": 6.840764331210191, | |
| "grad_norm": 0.7225663791930595, | |
| "learning_rate": 6.433091351086829e-08, | |
| "loss": 0.1938, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09248000383377075, | |
| "step": 5910, | |
| "valid_targets_mean": 3037.0, | |
| "valid_targets_min": 1928 | |
| }, | |
| { | |
| "epoch": 6.846554719166184, | |
| "grad_norm": 0.6427345921650035, | |
| "learning_rate": 5.97882956942053e-08, | |
| "loss": 0.1808, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0936291515827179, | |
| "step": 5915, | |
| "valid_targets_mean": 4733.2, | |
| "valid_targets_min": 3147 | |
| }, | |
| { | |
| "epoch": 6.8523451071221775, | |
| "grad_norm": 0.652378535416869, | |
| "learning_rate": 5.541174839883256e-08, | |
| "loss": 0.1865, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08591797947883606, | |
| "step": 5920, | |
| "valid_targets_mean": 3348.4, | |
| "valid_targets_min": 1844 | |
| }, | |
| { | |
| "epoch": 6.85813549507817, | |
| "grad_norm": 0.7084978224727022, | |
| "learning_rate": 5.12013080744822e-08, | |
| "loss": 0.1891, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09425882995128632, | |
| "step": 5925, | |
| "valid_targets_mean": 4360.2, | |
| "valid_targets_min": 2330 | |
| }, | |
| { | |
| "epoch": 6.863925883034163, | |
| "grad_norm": 0.7953115356519015, | |
| "learning_rate": 4.7157009787486276e-08, | |
| "loss": 0.1965, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10636697709560394, | |
| "step": 5930, | |
| "valid_targets_mean": 3574.1, | |
| "valid_targets_min": 900 | |
| }, | |
| { | |
| "epoch": 6.869716270990156, | |
| "grad_norm": 0.6314888078450656, | |
| "learning_rate": 4.3278887220470355e-08, | |
| "loss": 0.1875, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10081933438777924, | |
| "step": 5935, | |
| "valid_targets_mean": 5092.8, | |
| "valid_targets_min": 1713 | |
| }, | |
| { | |
| "epoch": 6.87550665894615, | |
| "grad_norm": 0.7190545537104681, | |
| "learning_rate": 3.9566972672080427e-08, | |
| "loss": 0.1959, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10799182206392288, | |
| "step": 5940, | |
| "valid_targets_mean": 4057.1, | |
| "valid_targets_min": 981 | |
| }, | |
| { | |
| "epoch": 6.881297046902143, | |
| "grad_norm": 0.6706026201273239, | |
| "learning_rate": 3.60212970567142e-08, | |
| "loss": 0.1922, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11171825975179672, | |
| "step": 5945, | |
| "valid_targets_mean": 4127.9, | |
| "valid_targets_min": 2098 | |
| }, | |
| { | |
| "epoch": 6.8870874348581355, | |
| "grad_norm": 0.7490324494092551, | |
| "learning_rate": 3.2641889904256875e-08, | |
| "loss": 0.1915, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08943615853786469, | |
| "step": 5950, | |
| "valid_targets_mean": 3101.1, | |
| "valid_targets_min": 1107 | |
| }, | |
| { | |
| "epoch": 6.892877822814128, | |
| "grad_norm": 0.6772633245975402, | |
| "learning_rate": 2.942877935983912e-08, | |
| "loss": 0.1922, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09332503378391266, | |
| "step": 5955, | |
| "valid_targets_mean": 3718.0, | |
| "valid_targets_min": 2273 | |
| }, | |
| { | |
| "epoch": 6.898668210770122, | |
| "grad_norm": 0.70928581545397, | |
| "learning_rate": 2.6381992183610593e-08, | |
| "loss": 0.1906, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08233386278152466, | |
| "step": 5960, | |
| "valid_targets_mean": 3316.0, | |
| "valid_targets_min": 1920 | |
| }, | |
| { | |
| "epoch": 6.904458598726115, | |
| "grad_norm": 1.4468656575646994, | |
| "learning_rate": 2.3501553750497897e-08, | |
| "loss": 0.1782, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09502074867486954, | |
| "step": 5965, | |
| "valid_targets_mean": 4710.8, | |
| "valid_targets_min": 1687 | |
| }, | |
| { | |
| "epoch": 6.910248986682108, | |
| "grad_norm": 0.7764638334594126, | |
| "learning_rate": 2.0787488050015846e-08, | |
| "loss": 0.1855, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08005072921514511, | |
| "step": 5970, | |
| "valid_targets_mean": 3075.8, | |
| "valid_targets_min": 1375 | |
| }, | |
| { | |
| "epoch": 6.916039374638101, | |
| "grad_norm": 0.7063658629510593, | |
| "learning_rate": 1.8239817686049877e-08, | |
| "loss": 0.1873, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08526700735092163, | |
| "step": 5975, | |
| "valid_targets_mean": 3776.4, | |
| "valid_targets_min": 906 | |
| }, | |
| { | |
| "epoch": 6.921829762594093, | |
| "grad_norm": 0.7338832979330177, | |
| "learning_rate": 1.585856387667839e-08, | |
| "loss": 0.1866, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09271879494190216, | |
| "step": 5980, | |
| "valid_targets_mean": 3393.9, | |
| "valid_targets_min": 1874 | |
| }, | |
| { | |
| "epoch": 6.927620150550087, | |
| "grad_norm": 0.6875599163112425, | |
| "learning_rate": 1.3643746453990692e-08, | |
| "loss": 0.1791, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09227672219276428, | |
| "step": 5985, | |
| "valid_targets_mean": 3503.4, | |
| "valid_targets_min": 1877 | |
| }, | |
| { | |
| "epoch": 6.93341053850608, | |
| "grad_norm": 0.6880943124412784, | |
| "learning_rate": 1.1595383863922672e-08, | |
| "loss": 0.1832, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09128659963607788, | |
| "step": 5990, | |
| "valid_targets_mean": 3744.4, | |
| "valid_targets_min": 1697 | |
| }, | |
| { | |
| "epoch": 6.939200926462073, | |
| "grad_norm": 0.7890699686552586, | |
| "learning_rate": 9.713493166105814e-09, | |
| "loss": 0.1888, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10081125050783157, | |
| "step": 5995, | |
| "valid_targets_mean": 3015.1, | |
| "valid_targets_min": 1679 | |
| }, | |
| { | |
| "epoch": 6.944991314418066, | |
| "grad_norm": 0.7467220344761751, | |
| "learning_rate": 7.998090033720652e-09, | |
| "loss": 0.2055, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10079054534435272, | |
| "step": 6000, | |
| "valid_targets_mean": 3292.0, | |
| "valid_targets_min": 1695 | |
| }, | |
| { | |
| "epoch": 6.950781702374059, | |
| "grad_norm": 0.7100018820654072, | |
| "learning_rate": 6.4491887533701945e-09, | |
| "loss": 0.19, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08281861245632172, | |
| "step": 6005, | |
| "valid_targets_mean": 3148.8, | |
| "valid_targets_min": 2114 | |
| }, | |
| { | |
| "epoch": 6.956572090330052, | |
| "grad_norm": 0.7037414374916773, | |
| "learning_rate": 5.066802224962253e-09, | |
| "loss": 0.1848, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07482528686523438, | |
| "step": 6010, | |
| "valid_targets_mean": 2852.0, | |
| "valid_targets_min": 993 | |
| }, | |
| { | |
| "epoch": 6.962362478286045, | |
| "grad_norm": 0.7469951851474523, | |
| "learning_rate": 3.850941961593968e-09, | |
| "loss": 0.1808, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08902069926261902, | |
| "step": 6015, | |
| "valid_targets_mean": 3059.8, | |
| "valid_targets_min": 1275 | |
| }, | |
| { | |
| "epoch": 6.968152866242038, | |
| "grad_norm": 0.6463801645047006, | |
| "learning_rate": 2.801618089465219e-09, | |
| "loss": 0.1931, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10158857703208923, | |
| "step": 6020, | |
| "valid_targets_mean": 4479.0, | |
| "valid_targets_min": 2250 | |
| }, | |
| { | |
| "epoch": 6.973943254198032, | |
| "grad_norm": 0.7019589980202549, | |
| "learning_rate": 1.918839347792023e-09, | |
| "loss": 0.1888, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11621302366256714, | |
| "step": 6025, | |
| "valid_targets_mean": 4952.2, | |
| "valid_targets_min": 2892 | |
| }, | |
| { | |
| "epoch": 6.979733642154025, | |
| "grad_norm": 0.6735471715415253, | |
| "learning_rate": 1.2026130887243803e-09, | |
| "loss": 0.1969, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.0936989039182663, | |
| "step": 6030, | |
| "valid_targets_mean": 4026.4, | |
| "valid_targets_min": 2935 | |
| }, | |
| { | |
| "epoch": 6.985524030110017, | |
| "grad_norm": 0.678477506458588, | |
| "learning_rate": 6.529452773018641e-10, | |
| "loss": 0.1908, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.08855093270540237, | |
| "step": 6035, | |
| "valid_targets_mean": 4085.2, | |
| "valid_targets_min": 1817 | |
| }, | |
| { | |
| "epoch": 6.99131441806601, | |
| "grad_norm": 0.7191713008057828, | |
| "learning_rate": 2.6984049138700785e-10, | |
| "loss": 0.1896, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10044793039560318, | |
| "step": 6040, | |
| "valid_targets_mean": 4050.4, | |
| "valid_targets_min": 1727 | |
| }, | |
| { | |
| "epoch": 6.997104806022003, | |
| "grad_norm": 0.6412694764109674, | |
| "learning_rate": 5.330192164310077e-11, | |
| "loss": 0.1906, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.07490961253643036, | |
| "step": 6045, | |
| "valid_targets_mean": 3682.9, | |
| "valid_targets_min": 1622 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15162084996700287, | |
| "step": 6048, | |
| "total_flos": 2.4439457607529267e+18, | |
| "train_loss": 0.23659130134595135, | |
| "train_runtime": 153851.6493, | |
| "train_samples_per_second": 0.628, | |
| "train_steps_per_second": 0.039, | |
| "valid_targets_mean": 5459.5, | |
| "valid_targets_min": 2092 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 6048, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 7, | |
| "save_steps": 200, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2.4439457607529267e+18, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |