Text Generation
Transformers
Safetensors
qwen3
llama-factory
full
Generated from Trainer
conversational
text-generation-inference
Instructions to use DCAgent/a1-nemotron_cpp with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use DCAgent/a1-nemotron_cpp with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="DCAgent/a1-nemotron_cpp") messages = [ {"role": "user", "content": "Who are you?"}, ] pipe(messages)# Load model directly from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("DCAgent/a1-nemotron_cpp") model = AutoModelForCausalLM.from_pretrained("DCAgent/a1-nemotron_cpp", 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 DCAgent/a1-nemotron_cpp with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "DCAgent/a1-nemotron_cpp" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "DCAgent/a1-nemotron_cpp", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/DCAgent/a1-nemotron_cpp
- SGLang
How to use DCAgent/a1-nemotron_cpp 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 "DCAgent/a1-nemotron_cpp" \ --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": "DCAgent/a1-nemotron_cpp", "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 "DCAgent/a1-nemotron_cpp" \ --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": "DCAgent/a1-nemotron_cpp", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }' - Docker Model Runner
How to use DCAgent/a1-nemotron_cpp with Docker Model Runner:
docker model run hf.co/DCAgent/a1-nemotron_cpp
| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 7.0, | |
| "eval_steps": 500, | |
| "global_step": 4382, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.007987220447284345, | |
| "grad_norm": 14.750937141242062, | |
| "learning_rate": 3.644646924829157e-07, | |
| "loss": 0.6181, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.5868211984634399, | |
| "step": 5, | |
| "valid_targets_mean": 5183.6, | |
| "valid_targets_min": 2026 | |
| }, | |
| { | |
| "epoch": 0.01597444089456869, | |
| "grad_norm": 14.964261167947019, | |
| "learning_rate": 8.200455580865605e-07, | |
| "loss": 0.6115, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.600242555141449, | |
| "step": 10, | |
| "valid_targets_mean": 4833.1, | |
| "valid_targets_min": 2416 | |
| }, | |
| { | |
| "epoch": 0.023961661341853034, | |
| "grad_norm": 14.204887464722148, | |
| "learning_rate": 1.2756264236902052e-06, | |
| "loss": 0.59, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.5956233143806458, | |
| "step": 15, | |
| "valid_targets_mean": 5260.3, | |
| "valid_targets_min": 2260 | |
| }, | |
| { | |
| "epoch": 0.03194888178913738, | |
| "grad_norm": 12.990536880380345, | |
| "learning_rate": 1.7312072892938498e-06, | |
| "loss": 0.6008, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.6236944198608398, | |
| "step": 20, | |
| "valid_targets_mean": 4872.7, | |
| "valid_targets_min": 2443 | |
| }, | |
| { | |
| "epoch": 0.039936102236421724, | |
| "grad_norm": 7.212127023695072, | |
| "learning_rate": 2.1867881548974945e-06, | |
| "loss": 0.5413, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.5282974243164062, | |
| "step": 25, | |
| "valid_targets_mean": 5652.8, | |
| "valid_targets_min": 3266 | |
| }, | |
| { | |
| "epoch": 0.04792332268370607, | |
| "grad_norm": 4.348281336433791, | |
| "learning_rate": 2.642369020501139e-06, | |
| "loss": 0.4748, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.42746517062187195, | |
| "step": 30, | |
| "valid_targets_mean": 5700.2, | |
| "valid_targets_min": 3211 | |
| }, | |
| { | |
| "epoch": 0.05591054313099041, | |
| "grad_norm": 2.60657809706934, | |
| "learning_rate": 3.0979498861047843e-06, | |
| "loss": 0.4258, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3874906003475189, | |
| "step": 35, | |
| "valid_targets_mean": 5090.2, | |
| "valid_targets_min": 2305 | |
| }, | |
| { | |
| "epoch": 0.06389776357827476, | |
| "grad_norm": 1.476512144338404, | |
| "learning_rate": 3.5535307517084285e-06, | |
| "loss": 0.4243, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3955470323562622, | |
| "step": 40, | |
| "valid_targets_mean": 5426.3, | |
| "valid_targets_min": 2239 | |
| }, | |
| { | |
| "epoch": 0.07188498402555911, | |
| "grad_norm": 1.101318544550606, | |
| "learning_rate": 4.009111617312073e-06, | |
| "loss": 0.3837, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3805215358734131, | |
| "step": 45, | |
| "valid_targets_mean": 5182.9, | |
| "valid_targets_min": 1585 | |
| }, | |
| { | |
| "epoch": 0.07987220447284345, | |
| "grad_norm": 1.018486806698862, | |
| "learning_rate": 4.464692482915718e-06, | |
| "loss": 0.384, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3917389512062073, | |
| "step": 50, | |
| "valid_targets_mean": 4319.9, | |
| "valid_targets_min": 1008 | |
| }, | |
| { | |
| "epoch": 0.0878594249201278, | |
| "grad_norm": 0.840115633146457, | |
| "learning_rate": 4.920273348519363e-06, | |
| "loss": 0.3666, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.38408344984054565, | |
| "step": 55, | |
| "valid_targets_mean": 5725.0, | |
| "valid_targets_min": 3444 | |
| }, | |
| { | |
| "epoch": 0.09584664536741214, | |
| "grad_norm": 0.6882080666422143, | |
| "learning_rate": 5.375854214123008e-06, | |
| "loss": 0.3544, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3829645812511444, | |
| "step": 60, | |
| "valid_targets_mean": 5371.8, | |
| "valid_targets_min": 2168 | |
| }, | |
| { | |
| "epoch": 0.10383386581469649, | |
| "grad_norm": 0.5884122222611426, | |
| "learning_rate": 5.831435079726651e-06, | |
| "loss": 0.3363, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2908532917499542, | |
| "step": 65, | |
| "valid_targets_mean": 5195.1, | |
| "valid_targets_min": 1604 | |
| }, | |
| { | |
| "epoch": 0.11182108626198083, | |
| "grad_norm": 0.6714646781973334, | |
| "learning_rate": 6.287015945330297e-06, | |
| "loss": 0.3449, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.35886451601982117, | |
| "step": 70, | |
| "valid_targets_mean": 4314.6, | |
| "valid_targets_min": 823 | |
| }, | |
| { | |
| "epoch": 0.11980830670926518, | |
| "grad_norm": 0.6524389753252751, | |
| "learning_rate": 6.742596810933942e-06, | |
| "loss": 0.3446, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3823692500591278, | |
| "step": 75, | |
| "valid_targets_mean": 5004.7, | |
| "valid_targets_min": 2712 | |
| }, | |
| { | |
| "epoch": 0.12779552715654952, | |
| "grad_norm": 0.44934600305386857, | |
| "learning_rate": 7.1981776765375854e-06, | |
| "loss": 0.3146, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2777388095855713, | |
| "step": 80, | |
| "valid_targets_mean": 5506.7, | |
| "valid_targets_min": 3003 | |
| }, | |
| { | |
| "epoch": 0.13578274760383385, | |
| "grad_norm": 0.5479356262430486, | |
| "learning_rate": 7.65375854214123e-06, | |
| "loss": 0.3123, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2883882224559784, | |
| "step": 85, | |
| "valid_targets_mean": 4666.4, | |
| "valid_targets_min": 868 | |
| }, | |
| { | |
| "epoch": 0.14376996805111822, | |
| "grad_norm": 0.5283663829799473, | |
| "learning_rate": 8.109339407744875e-06, | |
| "loss": 0.3167, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.30486780405044556, | |
| "step": 90, | |
| "valid_targets_mean": 5202.8, | |
| "valid_targets_min": 1489 | |
| }, | |
| { | |
| "epoch": 0.15175718849840256, | |
| "grad_norm": 0.48446883276952224, | |
| "learning_rate": 8.564920273348521e-06, | |
| "loss": 0.3141, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.286685049533844, | |
| "step": 95, | |
| "valid_targets_mean": 5233.1, | |
| "valid_targets_min": 2337 | |
| }, | |
| { | |
| "epoch": 0.1597444089456869, | |
| "grad_norm": 0.4951046449787275, | |
| "learning_rate": 9.020501138952164e-06, | |
| "loss": 0.3066, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27557373046875, | |
| "step": 100, | |
| "valid_targets_mean": 5019.7, | |
| "valid_targets_min": 1527 | |
| }, | |
| { | |
| "epoch": 0.16773162939297126, | |
| "grad_norm": 0.5397619382354931, | |
| "learning_rate": 9.47608200455581e-06, | |
| "loss": 0.2681, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2617327570915222, | |
| "step": 105, | |
| "valid_targets_mean": 4160.9, | |
| "valid_targets_min": 673 | |
| }, | |
| { | |
| "epoch": 0.1757188498402556, | |
| "grad_norm": 0.4594794804574554, | |
| "learning_rate": 9.931662870159453e-06, | |
| "loss": 0.3039, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26641079783439636, | |
| "step": 110, | |
| "valid_targets_mean": 5543.9, | |
| "valid_targets_min": 2528 | |
| }, | |
| { | |
| "epoch": 0.18370607028753994, | |
| "grad_norm": 0.4467064356678536, | |
| "learning_rate": 1.03872437357631e-05, | |
| "loss": 0.2895, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2728346884250641, | |
| "step": 115, | |
| "valid_targets_mean": 5978.6, | |
| "valid_targets_min": 2496 | |
| }, | |
| { | |
| "epoch": 0.19169329073482427, | |
| "grad_norm": 0.4909322598820612, | |
| "learning_rate": 1.0842824601366744e-05, | |
| "loss": 0.2844, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2716064453125, | |
| "step": 120, | |
| "valid_targets_mean": 4534.9, | |
| "valid_targets_min": 966 | |
| }, | |
| { | |
| "epoch": 0.19968051118210864, | |
| "grad_norm": 0.5280762038220295, | |
| "learning_rate": 1.1298405466970387e-05, | |
| "loss": 0.2855, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27411317825317383, | |
| "step": 125, | |
| "valid_targets_mean": 5088.2, | |
| "valid_targets_min": 1452 | |
| }, | |
| { | |
| "epoch": 0.20766773162939298, | |
| "grad_norm": 0.44322041012524493, | |
| "learning_rate": 1.1753986332574032e-05, | |
| "loss": 0.2729, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26688119769096375, | |
| "step": 130, | |
| "valid_targets_mean": 5735.3, | |
| "valid_targets_min": 3327 | |
| }, | |
| { | |
| "epoch": 0.21565495207667731, | |
| "grad_norm": 0.4768612654173803, | |
| "learning_rate": 1.2209567198177677e-05, | |
| "loss": 0.2726, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2971537709236145, | |
| "step": 135, | |
| "valid_targets_mean": 5256.0, | |
| "valid_targets_min": 1604 | |
| }, | |
| { | |
| "epoch": 0.22364217252396165, | |
| "grad_norm": 0.47239494063262155, | |
| "learning_rate": 1.2665148063781323e-05, | |
| "loss": 0.2631, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2521602213382721, | |
| "step": 140, | |
| "valid_targets_mean": 4993.7, | |
| "valid_targets_min": 1627 | |
| }, | |
| { | |
| "epoch": 0.23162939297124602, | |
| "grad_norm": 0.5243809619016284, | |
| "learning_rate": 1.3120728929384968e-05, | |
| "loss": 0.2632, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2900952994823456, | |
| "step": 145, | |
| "valid_targets_mean": 4655.7, | |
| "valid_targets_min": 1368 | |
| }, | |
| { | |
| "epoch": 0.23961661341853036, | |
| "grad_norm": 0.4940531962711237, | |
| "learning_rate": 1.357630979498861e-05, | |
| "loss": 0.2601, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28482481837272644, | |
| "step": 150, | |
| "valid_targets_mean": 4984.1, | |
| "valid_targets_min": 655 | |
| }, | |
| { | |
| "epoch": 0.2476038338658147, | |
| "grad_norm": 0.4898028743062861, | |
| "learning_rate": 1.4031890660592255e-05, | |
| "loss": 0.2652, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24906936287879944, | |
| "step": 155, | |
| "valid_targets_mean": 5235.4, | |
| "valid_targets_min": 2808 | |
| }, | |
| { | |
| "epoch": 0.25559105431309903, | |
| "grad_norm": 0.48837141721890953, | |
| "learning_rate": 1.4487471526195902e-05, | |
| "loss": 0.244, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2593861222267151, | |
| "step": 160, | |
| "valid_targets_mean": 5388.4, | |
| "valid_targets_min": 2425 | |
| }, | |
| { | |
| "epoch": 0.26357827476038337, | |
| "grad_norm": 0.568541877756852, | |
| "learning_rate": 1.4943052391799546e-05, | |
| "loss": 0.2478, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2541233003139496, | |
| "step": 165, | |
| "valid_targets_mean": 4213.4, | |
| "valid_targets_min": 1563 | |
| }, | |
| { | |
| "epoch": 0.2715654952076677, | |
| "grad_norm": 0.5284463940586939, | |
| "learning_rate": 1.539863325740319e-05, | |
| "loss": 0.2541, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2715235650539398, | |
| "step": 170, | |
| "valid_targets_mean": 4626.6, | |
| "valid_targets_min": 1065 | |
| }, | |
| { | |
| "epoch": 0.2795527156549521, | |
| "grad_norm": 0.4808680852503573, | |
| "learning_rate": 1.5854214123006836e-05, | |
| "loss": 0.2504, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27802127599716187, | |
| "step": 175, | |
| "valid_targets_mean": 5711.0, | |
| "valid_targets_min": 3891 | |
| }, | |
| { | |
| "epoch": 0.28753993610223644, | |
| "grad_norm": 0.48423409008710366, | |
| "learning_rate": 1.630979498861048e-05, | |
| "loss": 0.2471, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2705083191394806, | |
| "step": 180, | |
| "valid_targets_mean": 6178.1, | |
| "valid_targets_min": 1925 | |
| }, | |
| { | |
| "epoch": 0.2955271565495208, | |
| "grad_norm": 0.433249461013068, | |
| "learning_rate": 1.6765375854214125e-05, | |
| "loss": 0.2373, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22675535082817078, | |
| "step": 185, | |
| "valid_targets_mean": 5465.7, | |
| "valid_targets_min": 1838 | |
| }, | |
| { | |
| "epoch": 0.3035143769968051, | |
| "grad_norm": 0.561750111126097, | |
| "learning_rate": 1.722095671981777e-05, | |
| "loss": 0.247, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24159491062164307, | |
| "step": 190, | |
| "valid_targets_mean": 5817.1, | |
| "valid_targets_min": 2297 | |
| }, | |
| { | |
| "epoch": 0.31150159744408945, | |
| "grad_norm": 0.5167037717710051, | |
| "learning_rate": 1.7676537585421415e-05, | |
| "loss": 0.248, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.238520547747612, | |
| "step": 195, | |
| "valid_targets_mean": 5234.8, | |
| "valid_targets_min": 2731 | |
| }, | |
| { | |
| "epoch": 0.3194888178913738, | |
| "grad_norm": 0.5176567260652404, | |
| "learning_rate": 1.813211845102506e-05, | |
| "loss": 0.2351, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2186240404844284, | |
| "step": 200, | |
| "valid_targets_mean": 4947.4, | |
| "valid_targets_min": 2587 | |
| }, | |
| { | |
| "epoch": 0.3274760383386581, | |
| "grad_norm": 0.49839155972836263, | |
| "learning_rate": 1.8587699316628704e-05, | |
| "loss": 0.2425, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24029989540576935, | |
| "step": 205, | |
| "valid_targets_mean": 5392.4, | |
| "valid_targets_min": 3121 | |
| }, | |
| { | |
| "epoch": 0.3354632587859425, | |
| "grad_norm": 0.514546853522046, | |
| "learning_rate": 1.904328018223235e-05, | |
| "loss": 0.2485, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2890052795410156, | |
| "step": 210, | |
| "valid_targets_mean": 5127.9, | |
| "valid_targets_min": 1493 | |
| }, | |
| { | |
| "epoch": 0.34345047923322686, | |
| "grad_norm": 0.5398109526641716, | |
| "learning_rate": 1.9498861047835993e-05, | |
| "loss": 0.2354, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24977071583271027, | |
| "step": 215, | |
| "valid_targets_mean": 4957.1, | |
| "valid_targets_min": 1513 | |
| }, | |
| { | |
| "epoch": 0.3514376996805112, | |
| "grad_norm": 0.48890472915927236, | |
| "learning_rate": 1.9954441913439638e-05, | |
| "loss": 0.248, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25104936957359314, | |
| "step": 220, | |
| "valid_targets_mean": 5089.3, | |
| "valid_targets_min": 2779 | |
| }, | |
| { | |
| "epoch": 0.35942492012779553, | |
| "grad_norm": 0.5195930703265889, | |
| "learning_rate": 2.0410022779043283e-05, | |
| "loss": 0.2304, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22001594305038452, | |
| "step": 225, | |
| "valid_targets_mean": 4320.8, | |
| "valid_targets_min": 1448 | |
| }, | |
| { | |
| "epoch": 0.36741214057507987, | |
| "grad_norm": 0.49684037290127403, | |
| "learning_rate": 2.0865603644646927e-05, | |
| "loss": 0.2316, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2174071967601776, | |
| "step": 230, | |
| "valid_targets_mean": 4716.7, | |
| "valid_targets_min": 1571 | |
| }, | |
| { | |
| "epoch": 0.3753993610223642, | |
| "grad_norm": 0.5730133717608953, | |
| "learning_rate": 2.1321184510250572e-05, | |
| "loss": 0.2457, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21891537308692932, | |
| "step": 235, | |
| "valid_targets_mean": 4716.6, | |
| "valid_targets_min": 1383 | |
| }, | |
| { | |
| "epoch": 0.38338658146964855, | |
| "grad_norm": 0.5055909284483817, | |
| "learning_rate": 2.1776765375854217e-05, | |
| "loss": 0.2359, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22691525518894196, | |
| "step": 240, | |
| "valid_targets_mean": 5677.3, | |
| "valid_targets_min": 1871 | |
| }, | |
| { | |
| "epoch": 0.3913738019169329, | |
| "grad_norm": 0.4489454922539967, | |
| "learning_rate": 2.223234624145786e-05, | |
| "loss": 0.2338, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22592651844024658, | |
| "step": 245, | |
| "valid_targets_mean": 5471.2, | |
| "valid_targets_min": 3137 | |
| }, | |
| { | |
| "epoch": 0.3993610223642173, | |
| "grad_norm": 0.5540998394736883, | |
| "learning_rate": 2.2687927107061506e-05, | |
| "loss": 0.2342, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2527666687965393, | |
| "step": 250, | |
| "valid_targets_mean": 5065.4, | |
| "valid_targets_min": 563 | |
| }, | |
| { | |
| "epoch": 0.4073482428115016, | |
| "grad_norm": 0.4797512067448123, | |
| "learning_rate": 2.314350797266515e-05, | |
| "loss": 0.2299, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24020880460739136, | |
| "step": 255, | |
| "valid_targets_mean": 5891.6, | |
| "valid_targets_min": 3338 | |
| }, | |
| { | |
| "epoch": 0.41533546325878595, | |
| "grad_norm": 0.48887349171587857, | |
| "learning_rate": 2.3599088838268792e-05, | |
| "loss": 0.2204, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23238958418369293, | |
| "step": 260, | |
| "valid_targets_mean": 5263.6, | |
| "valid_targets_min": 2530 | |
| }, | |
| { | |
| "epoch": 0.4233226837060703, | |
| "grad_norm": 0.48455266052013457, | |
| "learning_rate": 2.4054669703872436e-05, | |
| "loss": 0.2197, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23205773532390594, | |
| "step": 265, | |
| "valid_targets_mean": 5051.2, | |
| "valid_targets_min": 2893 | |
| }, | |
| { | |
| "epoch": 0.43130990415335463, | |
| "grad_norm": 0.49785923139416033, | |
| "learning_rate": 2.4510250569476085e-05, | |
| "loss": 0.2391, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2300228774547577, | |
| "step": 270, | |
| "valid_targets_mean": 4779.6, | |
| "valid_targets_min": 2985 | |
| }, | |
| { | |
| "epoch": 0.43929712460063897, | |
| "grad_norm": 0.5673155256006802, | |
| "learning_rate": 2.496583143507973e-05, | |
| "loss": 0.2248, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22798585891723633, | |
| "step": 275, | |
| "valid_targets_mean": 5471.0, | |
| "valid_targets_min": 2601 | |
| }, | |
| { | |
| "epoch": 0.4472843450479233, | |
| "grad_norm": 0.49503898465929125, | |
| "learning_rate": 2.5421412300683374e-05, | |
| "loss": 0.2424, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2597437798976898, | |
| "step": 280, | |
| "valid_targets_mean": 5064.5, | |
| "valid_targets_min": 3127 | |
| }, | |
| { | |
| "epoch": 0.45527156549520764, | |
| "grad_norm": 0.4887141545058853, | |
| "learning_rate": 2.587699316628702e-05, | |
| "loss": 0.231, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2254042625427246, | |
| "step": 285, | |
| "valid_targets_mean": 5261.2, | |
| "valid_targets_min": 2456 | |
| }, | |
| { | |
| "epoch": 0.46325878594249204, | |
| "grad_norm": 0.5299539976203823, | |
| "learning_rate": 2.6332574031890663e-05, | |
| "loss": 0.2507, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23290681838989258, | |
| "step": 290, | |
| "valid_targets_mean": 5007.8, | |
| "valid_targets_min": 1811 | |
| }, | |
| { | |
| "epoch": 0.4712460063897764, | |
| "grad_norm": 0.4572874629579899, | |
| "learning_rate": 2.6788154897494308e-05, | |
| "loss": 0.2356, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18511193990707397, | |
| "step": 295, | |
| "valid_targets_mean": 5224.5, | |
| "valid_targets_min": 1799 | |
| }, | |
| { | |
| "epoch": 0.4792332268370607, | |
| "grad_norm": 0.45972574793046705, | |
| "learning_rate": 2.7243735763097953e-05, | |
| "loss": 0.2084, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19755448400974274, | |
| "step": 300, | |
| "valid_targets_mean": 5047.1, | |
| "valid_targets_min": 2289 | |
| }, | |
| { | |
| "epoch": 0.48722044728434505, | |
| "grad_norm": 0.5400550314354791, | |
| "learning_rate": 2.7699316628701597e-05, | |
| "loss": 0.2415, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21521860361099243, | |
| "step": 305, | |
| "valid_targets_mean": 5351.9, | |
| "valid_targets_min": 3167 | |
| }, | |
| { | |
| "epoch": 0.4952076677316294, | |
| "grad_norm": 0.4935745339384302, | |
| "learning_rate": 2.815489749430524e-05, | |
| "loss": 0.2128, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2429409623146057, | |
| "step": 310, | |
| "valid_targets_mean": 5082.1, | |
| "valid_targets_min": 1565 | |
| }, | |
| { | |
| "epoch": 0.5031948881789138, | |
| "grad_norm": 0.4938772146997711, | |
| "learning_rate": 2.8610478359908883e-05, | |
| "loss": 0.2131, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20227649807929993, | |
| "step": 315, | |
| "valid_targets_mean": 5009.8, | |
| "valid_targets_min": 1100 | |
| }, | |
| { | |
| "epoch": 0.5111821086261981, | |
| "grad_norm": 0.5800311114967133, | |
| "learning_rate": 2.906605922551253e-05, | |
| "loss": 0.2094, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19665858149528503, | |
| "step": 320, | |
| "valid_targets_mean": 5210.2, | |
| "valid_targets_min": 2165 | |
| }, | |
| { | |
| "epoch": 0.5191693290734825, | |
| "grad_norm": 0.47229462464182226, | |
| "learning_rate": 2.9521640091116176e-05, | |
| "loss": 0.2208, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22188128530979156, | |
| "step": 325, | |
| "valid_targets_mean": 5149.5, | |
| "valid_targets_min": 2528 | |
| }, | |
| { | |
| "epoch": 0.5271565495207667, | |
| "grad_norm": 0.4736617340620026, | |
| "learning_rate": 2.997722095671982e-05, | |
| "loss": 0.2161, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20165565609931946, | |
| "step": 330, | |
| "valid_targets_mean": 5324.2, | |
| "valid_targets_min": 1867 | |
| }, | |
| { | |
| "epoch": 0.5351437699680511, | |
| "grad_norm": 0.4435418248009787, | |
| "learning_rate": 3.0432801822323465e-05, | |
| "loss": 0.2227, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20023328065872192, | |
| "step": 335, | |
| "valid_targets_mean": 5795.4, | |
| "valid_targets_min": 3702 | |
| }, | |
| { | |
| "epoch": 0.5431309904153354, | |
| "grad_norm": 0.49528345004469515, | |
| "learning_rate": 3.088838268792711e-05, | |
| "loss": 0.2303, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22503232955932617, | |
| "step": 340, | |
| "valid_targets_mean": 4877.2, | |
| "valid_targets_min": 496 | |
| }, | |
| { | |
| "epoch": 0.5511182108626198, | |
| "grad_norm": 0.5187708531462577, | |
| "learning_rate": 3.1343963553530755e-05, | |
| "loss": 0.2263, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28174805641174316, | |
| "step": 345, | |
| "valid_targets_mean": 5624.0, | |
| "valid_targets_min": 3026 | |
| }, | |
| { | |
| "epoch": 0.5591054313099042, | |
| "grad_norm": 0.4903669062536284, | |
| "learning_rate": 3.17995444191344e-05, | |
| "loss": 0.2319, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2515377402305603, | |
| "step": 350, | |
| "valid_targets_mean": 5358.2, | |
| "valid_targets_min": 2463 | |
| }, | |
| { | |
| "epoch": 0.5670926517571885, | |
| "grad_norm": 0.6441307837885047, | |
| "learning_rate": 3.2255125284738044e-05, | |
| "loss": 0.2198, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22735492885112762, | |
| "step": 355, | |
| "valid_targets_mean": 5304.9, | |
| "valid_targets_min": 2203 | |
| }, | |
| { | |
| "epoch": 0.5750798722044729, | |
| "grad_norm": 0.4544546405060071, | |
| "learning_rate": 3.271070615034169e-05, | |
| "loss": 0.2225, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23330813646316528, | |
| "step": 360, | |
| "valid_targets_mean": 5305.2, | |
| "valid_targets_min": 2371 | |
| }, | |
| { | |
| "epoch": 0.5830670926517572, | |
| "grad_norm": 0.45087367913946724, | |
| "learning_rate": 3.316628701594533e-05, | |
| "loss": 0.2259, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.186964750289917, | |
| "step": 365, | |
| "valid_targets_mean": 5236.5, | |
| "valid_targets_min": 1953 | |
| }, | |
| { | |
| "epoch": 0.5910543130990416, | |
| "grad_norm": 0.5288590169436341, | |
| "learning_rate": 3.362186788154898e-05, | |
| "loss": 0.2242, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22872258722782135, | |
| "step": 370, | |
| "valid_targets_mean": 4543.8, | |
| "valid_targets_min": 2739 | |
| }, | |
| { | |
| "epoch": 0.5990415335463258, | |
| "grad_norm": 0.45707771597420815, | |
| "learning_rate": 3.407744874715262e-05, | |
| "loss": 0.2163, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20470118522644043, | |
| "step": 375, | |
| "valid_targets_mean": 5215.3, | |
| "valid_targets_min": 2614 | |
| }, | |
| { | |
| "epoch": 0.6070287539936102, | |
| "grad_norm": 0.48120517176319794, | |
| "learning_rate": 3.453302961275627e-05, | |
| "loss": 0.2206, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23557770252227783, | |
| "step": 380, | |
| "valid_targets_mean": 5652.8, | |
| "valid_targets_min": 2449 | |
| }, | |
| { | |
| "epoch": 0.6150159744408946, | |
| "grad_norm": 0.5548389036648828, | |
| "learning_rate": 3.498861047835991e-05, | |
| "loss": 0.2391, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23380829393863678, | |
| "step": 385, | |
| "valid_targets_mean": 4579.2, | |
| "valid_targets_min": 2715 | |
| }, | |
| { | |
| "epoch": 0.6230031948881789, | |
| "grad_norm": 0.47878863229908813, | |
| "learning_rate": 3.5444191343963557e-05, | |
| "loss": 0.2267, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24563977122306824, | |
| "step": 390, | |
| "valid_targets_mean": 5510.4, | |
| "valid_targets_min": 2402 | |
| }, | |
| { | |
| "epoch": 0.6309904153354633, | |
| "grad_norm": 0.42432757525496345, | |
| "learning_rate": 3.58997722095672e-05, | |
| "loss": 0.2269, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24582986533641815, | |
| "step": 395, | |
| "valid_targets_mean": 5569.0, | |
| "valid_targets_min": 2489 | |
| }, | |
| { | |
| "epoch": 0.6389776357827476, | |
| "grad_norm": 0.44057122961001555, | |
| "learning_rate": 3.6355353075170846e-05, | |
| "loss": 0.1985, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19309523701667786, | |
| "step": 400, | |
| "valid_targets_mean": 5235.2, | |
| "valid_targets_min": 2041 | |
| }, | |
| { | |
| "epoch": 0.646964856230032, | |
| "grad_norm": 0.4977064964494498, | |
| "learning_rate": 3.681093394077449e-05, | |
| "loss": 0.2135, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2071743607521057, | |
| "step": 405, | |
| "valid_targets_mean": 5551.8, | |
| "valid_targets_min": 3403 | |
| }, | |
| { | |
| "epoch": 0.6549520766773163, | |
| "grad_norm": 0.49716152713897765, | |
| "learning_rate": 3.7266514806378135e-05, | |
| "loss": 0.209, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22141389548778534, | |
| "step": 410, | |
| "valid_targets_mean": 4529.8, | |
| "valid_targets_min": 1310 | |
| }, | |
| { | |
| "epoch": 0.6629392971246006, | |
| "grad_norm": 0.49538859384179273, | |
| "learning_rate": 3.772209567198178e-05, | |
| "loss": 0.2167, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23128153383731842, | |
| "step": 415, | |
| "valid_targets_mean": 4873.6, | |
| "valid_targets_min": 973 | |
| }, | |
| { | |
| "epoch": 0.670926517571885, | |
| "grad_norm": 0.48057998955392545, | |
| "learning_rate": 3.8177676537585425e-05, | |
| "loss": 0.2203, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2080823928117752, | |
| "step": 420, | |
| "valid_targets_mean": 5251.4, | |
| "valid_targets_min": 3212 | |
| }, | |
| { | |
| "epoch": 0.6789137380191693, | |
| "grad_norm": 0.4839464587983944, | |
| "learning_rate": 3.863325740318907e-05, | |
| "loss": 0.2412, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2215818613767624, | |
| "step": 425, | |
| "valid_targets_mean": 5301.1, | |
| "valid_targets_min": 2760 | |
| }, | |
| { | |
| "epoch": 0.6869009584664537, | |
| "grad_norm": 0.46845217471033657, | |
| "learning_rate": 3.9088838268792714e-05, | |
| "loss": 0.228, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2471550554037094, | |
| "step": 430, | |
| "valid_targets_mean": 5756.3, | |
| "valid_targets_min": 3633 | |
| }, | |
| { | |
| "epoch": 0.694888178913738, | |
| "grad_norm": 0.5077755771178428, | |
| "learning_rate": 3.954441913439636e-05, | |
| "loss": 0.2182, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2147277593612671, | |
| "step": 435, | |
| "valid_targets_mean": 4530.1, | |
| "valid_targets_min": 434 | |
| }, | |
| { | |
| "epoch": 0.7028753993610224, | |
| "grad_norm": 0.5499766906688663, | |
| "learning_rate": 4e-05, | |
| "loss": 0.2194, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2366044968366623, | |
| "step": 440, | |
| "valid_targets_mean": 4289.3, | |
| "valid_targets_min": 967 | |
| }, | |
| { | |
| "epoch": 0.7108626198083067, | |
| "grad_norm": 0.4433266817131124, | |
| "learning_rate": 3.999984129682125e-05, | |
| "loss": 0.2119, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19199886918067932, | |
| "step": 445, | |
| "valid_targets_mean": 5310.1, | |
| "valid_targets_min": 2121 | |
| }, | |
| { | |
| "epoch": 0.7188498402555911, | |
| "grad_norm": 0.5599321784517268, | |
| "learning_rate": 3.9999365189803684e-05, | |
| "loss": 0.2053, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18982094526290894, | |
| "step": 450, | |
| "valid_targets_mean": 5324.1, | |
| "valid_targets_min": 2904 | |
| }, | |
| { | |
| "epoch": 0.7268370607028753, | |
| "grad_norm": 0.44161566701162475, | |
| "learning_rate": 3.9998571686503264e-05, | |
| "loss": 0.2108, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24004004895687103, | |
| "step": 455, | |
| "valid_targets_mean": 5432.5, | |
| "valid_targets_min": 1133 | |
| }, | |
| { | |
| "epoch": 0.7348242811501597, | |
| "grad_norm": 0.5194124076587047, | |
| "learning_rate": 3.9997460799513134e-05, | |
| "loss": 0.2269, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2347797453403473, | |
| "step": 460, | |
| "valid_targets_mean": 5217.9, | |
| "valid_targets_min": 1129 | |
| }, | |
| { | |
| "epoch": 0.7428115015974441, | |
| "grad_norm": 0.476648028191984, | |
| "learning_rate": 3.999603254646343e-05, | |
| "loss": 0.2109, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20515500009059906, | |
| "step": 465, | |
| "valid_targets_mean": 4342.2, | |
| "valid_targets_min": 1505 | |
| }, | |
| { | |
| "epoch": 0.7507987220447284, | |
| "grad_norm": 0.49610700962701476, | |
| "learning_rate": 3.9994286950020986e-05, | |
| "loss": 0.2161, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20888805389404297, | |
| "step": 470, | |
| "valid_targets_mean": 4414.0, | |
| "valid_targets_min": 1180 | |
| }, | |
| { | |
| "epoch": 0.7587859424920128, | |
| "grad_norm": 0.43750045351277805, | |
| "learning_rate": 3.999222403788896e-05, | |
| "loss": 0.21, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1941574513912201, | |
| "step": 475, | |
| "valid_targets_mean": 5818.1, | |
| "valid_targets_min": 3005 | |
| }, | |
| { | |
| "epoch": 0.7667731629392971, | |
| "grad_norm": 0.4347921164806212, | |
| "learning_rate": 3.9989843842806435e-05, | |
| "loss": 0.2157, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2178964912891388, | |
| "step": 480, | |
| "valid_targets_mean": 4973.9, | |
| "valid_targets_min": 967 | |
| }, | |
| { | |
| "epoch": 0.7747603833865815, | |
| "grad_norm": 0.4946860571603472, | |
| "learning_rate": 3.998714640254786e-05, | |
| "loss": 0.2212, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26357173919677734, | |
| "step": 485, | |
| "valid_targets_mean": 6071.4, | |
| "valid_targets_min": 2882 | |
| }, | |
| { | |
| "epoch": 0.7827476038338658, | |
| "grad_norm": 0.4422472166986323, | |
| "learning_rate": 3.998413175992247e-05, | |
| "loss": 0.2286, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22780314087867737, | |
| "step": 490, | |
| "valid_targets_mean": 5067.8, | |
| "valid_targets_min": 1698 | |
| }, | |
| { | |
| "epoch": 0.7907348242811502, | |
| "grad_norm": 0.46749132746767547, | |
| "learning_rate": 3.99807999627736e-05, | |
| "loss": 0.2157, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21358351409435272, | |
| "step": 495, | |
| "valid_targets_mean": 5144.9, | |
| "valid_targets_min": 895 | |
| }, | |
| { | |
| "epoch": 0.7987220447284346, | |
| "grad_norm": 0.41171595183574516, | |
| "learning_rate": 3.997715106397794e-05, | |
| "loss": 0.2266, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21357737481594086, | |
| "step": 500, | |
| "valid_targets_mean": 5765.4, | |
| "valid_targets_min": 3059 | |
| }, | |
| { | |
| "epoch": 0.8067092651757188, | |
| "grad_norm": 0.44379813000904833, | |
| "learning_rate": 3.997318512144465e-05, | |
| "loss": 0.2089, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23357757925987244, | |
| "step": 505, | |
| "valid_targets_mean": 5076.7, | |
| "valid_targets_min": 1880 | |
| }, | |
| { | |
| "epoch": 0.8146964856230032, | |
| "grad_norm": 0.36956736568418413, | |
| "learning_rate": 3.9968902198114516e-05, | |
| "loss": 0.2034, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1638990044593811, | |
| "step": 510, | |
| "valid_targets_mean": 5692.9, | |
| "valid_targets_min": 2659 | |
| }, | |
| { | |
| "epoch": 0.8226837060702875, | |
| "grad_norm": 0.4805683975900466, | |
| "learning_rate": 3.996430236195889e-05, | |
| "loss": 0.2257, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21078652143478394, | |
| "step": 515, | |
| "valid_targets_mean": 5165.9, | |
| "valid_targets_min": 644 | |
| }, | |
| { | |
| "epoch": 0.8306709265175719, | |
| "grad_norm": 0.4149574394098917, | |
| "learning_rate": 3.995938568597864e-05, | |
| "loss": 0.2042, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22348102927207947, | |
| "step": 520, | |
| "valid_targets_mean": 5319.1, | |
| "valid_targets_min": 2551 | |
| }, | |
| { | |
| "epoch": 0.8386581469648562, | |
| "grad_norm": 0.48694425442101597, | |
| "learning_rate": 3.995415224820297e-05, | |
| "loss": 0.2107, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16895313560962677, | |
| "step": 525, | |
| "valid_targets_mean": 5040.1, | |
| "valid_targets_min": 1514 | |
| }, | |
| { | |
| "epoch": 0.8466453674121406, | |
| "grad_norm": 0.43761034520250164, | |
| "learning_rate": 3.994860213168819e-05, | |
| "loss": 0.2043, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18031556904315948, | |
| "step": 530, | |
| "valid_targets_mean": 4929.1, | |
| "valid_targets_min": 2304 | |
| }, | |
| { | |
| "epoch": 0.854632587859425, | |
| "grad_norm": 0.46639254881988784, | |
| "learning_rate": 3.9942735424516435e-05, | |
| "loss": 0.21, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22063472867012024, | |
| "step": 535, | |
| "valid_targets_mean": 5225.8, | |
| "valid_targets_min": 882 | |
| }, | |
| { | |
| "epoch": 0.8626198083067093, | |
| "grad_norm": 0.4485848258179116, | |
| "learning_rate": 3.9936552219794196e-05, | |
| "loss": 0.2023, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19248777627944946, | |
| "step": 540, | |
| "valid_targets_mean": 4811.3, | |
| "valid_targets_min": 1618 | |
| }, | |
| { | |
| "epoch": 0.8706070287539937, | |
| "grad_norm": 0.4281925770385241, | |
| "learning_rate": 3.993005261565091e-05, | |
| "loss": 0.209, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1964084804058075, | |
| "step": 545, | |
| "valid_targets_mean": 5819.9, | |
| "valid_targets_min": 2591 | |
| }, | |
| { | |
| "epoch": 0.8785942492012779, | |
| "grad_norm": 0.47509103367289895, | |
| "learning_rate": 3.992323671523735e-05, | |
| "loss": 0.2101, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20187339186668396, | |
| "step": 550, | |
| "valid_targets_mean": 4858.6, | |
| "valid_targets_min": 1619 | |
| }, | |
| { | |
| "epoch": 0.8865814696485623, | |
| "grad_norm": 0.41513692917541734, | |
| "learning_rate": 3.991610462672403e-05, | |
| "loss": 0.2269, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21414723992347717, | |
| "step": 555, | |
| "valid_targets_mean": 5595.6, | |
| "valid_targets_min": 3144 | |
| }, | |
| { | |
| "epoch": 0.8945686900958466, | |
| "grad_norm": 0.4831646467414668, | |
| "learning_rate": 3.9908656463299456e-05, | |
| "loss": 0.1941, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18814024329185486, | |
| "step": 560, | |
| "valid_targets_mean": 4739.1, | |
| "valid_targets_min": 2593 | |
| }, | |
| { | |
| "epoch": 0.902555910543131, | |
| "grad_norm": 0.5585272737203633, | |
| "learning_rate": 3.990089234316835e-05, | |
| "loss": 0.2163, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26935335993766785, | |
| "step": 565, | |
| "valid_targets_mean": 5430.6, | |
| "valid_targets_min": 1253 | |
| }, | |
| { | |
| "epoch": 0.9105431309904153, | |
| "grad_norm": 0.4639941717016405, | |
| "learning_rate": 3.989281238954978e-05, | |
| "loss": 0.2055, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1973074972629547, | |
| "step": 570, | |
| "valid_targets_mean": 5237.6, | |
| "valid_targets_min": 2595 | |
| }, | |
| { | |
| "epoch": 0.9185303514376997, | |
| "grad_norm": 0.44096137390414514, | |
| "learning_rate": 3.9884416730675155e-05, | |
| "loss": 0.2109, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19466814398765564, | |
| "step": 575, | |
| "valid_targets_mean": 4507.9, | |
| "valid_targets_min": 2442 | |
| }, | |
| { | |
| "epoch": 0.9265175718849841, | |
| "grad_norm": 0.41668887411623967, | |
| "learning_rate": 3.987570549978626e-05, | |
| "loss": 0.2053, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2083684504032135, | |
| "step": 580, | |
| "valid_targets_mean": 5825.5, | |
| "valid_targets_min": 2557 | |
| }, | |
| { | |
| "epoch": 0.9345047923322684, | |
| "grad_norm": 0.4492482702869952, | |
| "learning_rate": 3.986667883513311e-05, | |
| "loss": 0.2035, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1946360021829605, | |
| "step": 585, | |
| "valid_targets_mean": 4618.2, | |
| "valid_targets_min": 1390 | |
| }, | |
| { | |
| "epoch": 0.9424920127795527, | |
| "grad_norm": 0.46526766237941025, | |
| "learning_rate": 3.985733687997173e-05, | |
| "loss": 0.201, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2234271913766861, | |
| "step": 590, | |
| "valid_targets_mean": 4822.6, | |
| "valid_targets_min": 2245 | |
| }, | |
| { | |
| "epoch": 0.950479233226837, | |
| "grad_norm": 0.41106012632882677, | |
| "learning_rate": 3.984767978256192e-05, | |
| "loss": 0.2053, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19287875294685364, | |
| "step": 595, | |
| "valid_targets_mean": 5341.6, | |
| "valid_targets_min": 1968 | |
| }, | |
| { | |
| "epoch": 0.9584664536741214, | |
| "grad_norm": 0.506999902043298, | |
| "learning_rate": 3.983770769616488e-05, | |
| "loss": 0.1982, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19591520726680756, | |
| "step": 600, | |
| "valid_targets_mean": 4244.9, | |
| "valid_targets_min": 2073 | |
| }, | |
| { | |
| "epoch": 0.9664536741214057, | |
| "grad_norm": 0.40051547357602163, | |
| "learning_rate": 3.9827420779040805e-05, | |
| "loss": 0.1959, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1894240826368332, | |
| "step": 605, | |
| "valid_targets_mean": 5670.6, | |
| "valid_targets_min": 3880 | |
| }, | |
| { | |
| "epoch": 0.9744408945686901, | |
| "grad_norm": 0.4560577254929694, | |
| "learning_rate": 3.981681919444633e-05, | |
| "loss": 0.2113, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2273043543100357, | |
| "step": 610, | |
| "valid_targets_mean": 4870.1, | |
| "valid_targets_min": 2808 | |
| }, | |
| { | |
| "epoch": 0.9824281150159745, | |
| "grad_norm": 0.4381902059885231, | |
| "learning_rate": 3.980590311063197e-05, | |
| "loss": 0.2079, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2070448100566864, | |
| "step": 615, | |
| "valid_targets_mean": 4721.6, | |
| "valid_targets_min": 437 | |
| }, | |
| { | |
| "epoch": 0.9904153354632588, | |
| "grad_norm": 0.5027222176700864, | |
| "learning_rate": 3.9794672700839455e-05, | |
| "loss": 0.2155, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20978277921676636, | |
| "step": 620, | |
| "valid_targets_mean": 4496.6, | |
| "valid_targets_min": 1431 | |
| }, | |
| { | |
| "epoch": 0.9984025559105432, | |
| "grad_norm": 0.462305999117298, | |
| "learning_rate": 3.9783128143298945e-05, | |
| "loss": 0.2201, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2238769680261612, | |
| "step": 625, | |
| "valid_targets_mean": 5288.6, | |
| "valid_targets_min": 1432 | |
| }, | |
| { | |
| "epoch": 1.0063897763578276, | |
| "grad_norm": 0.4306345684083587, | |
| "learning_rate": 3.977126962122625e-05, | |
| "loss": 0.1908, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17747753858566284, | |
| "step": 630, | |
| "valid_targets_mean": 4680.5, | |
| "valid_targets_min": 868 | |
| }, | |
| { | |
| "epoch": 1.0143769968051117, | |
| "grad_norm": 0.46067354091083634, | |
| "learning_rate": 3.975909732281988e-05, | |
| "loss": 0.2132, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22177430987358093, | |
| "step": 635, | |
| "valid_targets_mean": 4746.9, | |
| "valid_targets_min": 721 | |
| }, | |
| { | |
| "epoch": 1.0223642172523961, | |
| "grad_norm": 0.4090176597077107, | |
| "learning_rate": 3.974661144125808e-05, | |
| "loss": 0.1898, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16201627254486084, | |
| "step": 640, | |
| "valid_targets_mean": 5036.8, | |
| "valid_targets_min": 1180 | |
| }, | |
| { | |
| "epoch": 1.0303514376996805, | |
| "grad_norm": 0.4659876499906489, | |
| "learning_rate": 3.973381217469576e-05, | |
| "loss": 0.189, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19412903487682343, | |
| "step": 645, | |
| "valid_targets_mean": 5113.9, | |
| "valid_targets_min": 3079 | |
| }, | |
| { | |
| "epoch": 1.038338658146965, | |
| "grad_norm": 0.4372928459976877, | |
| "learning_rate": 3.972069972626135e-05, | |
| "loss": 0.1926, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1999775469303131, | |
| "step": 650, | |
| "valid_targets_mean": 5004.8, | |
| "valid_targets_min": 1515 | |
| }, | |
| { | |
| "epoch": 1.0463258785942493, | |
| "grad_norm": 0.41502100671309244, | |
| "learning_rate": 3.970727430405357e-05, | |
| "loss": 0.2073, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2547432780265808, | |
| "step": 655, | |
| "valid_targets_mean": 5895.9, | |
| "valid_targets_min": 2914 | |
| }, | |
| { | |
| "epoch": 1.0543130990415335, | |
| "grad_norm": 0.46327515569643063, | |
| "learning_rate": 3.969353612113815e-05, | |
| "loss": 0.2016, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21973958611488342, | |
| "step": 660, | |
| "valid_targets_mean": 4917.9, | |
| "valid_targets_min": 582 | |
| }, | |
| { | |
| "epoch": 1.0623003194888179, | |
| "grad_norm": 0.47684773073568687, | |
| "learning_rate": 3.96794853955444e-05, | |
| "loss": 0.197, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18203385174274445, | |
| "step": 665, | |
| "valid_targets_mean": 4358.0, | |
| "valid_targets_min": 300 | |
| }, | |
| { | |
| "epoch": 1.0702875399361023, | |
| "grad_norm": 0.5456060847806535, | |
| "learning_rate": 3.966512235026182e-05, | |
| "loss": 0.1847, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1914018839597702, | |
| "step": 670, | |
| "valid_targets_mean": 5252.8, | |
| "valid_targets_min": 2269 | |
| }, | |
| { | |
| "epoch": 1.0782747603833867, | |
| "grad_norm": 0.44575951218432336, | |
| "learning_rate": 3.96504472132365e-05, | |
| "loss": 0.2028, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17769761383533478, | |
| "step": 675, | |
| "valid_targets_mean": 4936.2, | |
| "valid_targets_min": 2705 | |
| }, | |
| { | |
| "epoch": 1.0862619808306708, | |
| "grad_norm": 0.40347381394995796, | |
| "learning_rate": 3.9635460217367513e-05, | |
| "loss": 0.1859, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17873942852020264, | |
| "step": 680, | |
| "valid_targets_mean": 5264.6, | |
| "valid_targets_min": 1736 | |
| }, | |
| { | |
| "epoch": 1.0942492012779552, | |
| "grad_norm": 0.4878271923566804, | |
| "learning_rate": 3.962016160050327e-05, | |
| "loss": 0.1884, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23434291779994965, | |
| "step": 685, | |
| "valid_targets_mean": 4679.9, | |
| "valid_targets_min": 2363 | |
| }, | |
| { | |
| "epoch": 1.1022364217252396, | |
| "grad_norm": 0.5245721396444153, | |
| "learning_rate": 3.960455160543767e-05, | |
| "loss": 0.1886, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19156387448310852, | |
| "step": 690, | |
| "valid_targets_mean": 5358.0, | |
| "valid_targets_min": 3113 | |
| }, | |
| { | |
| "epoch": 1.110223642172524, | |
| "grad_norm": 1.0046872184676034, | |
| "learning_rate": 3.958863047990631e-05, | |
| "loss": 0.2031, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22785481810569763, | |
| "step": 695, | |
| "valid_targets_mean": 5206.9, | |
| "valid_targets_min": 982 | |
| }, | |
| { | |
| "epoch": 1.1182108626198084, | |
| "grad_norm": 0.4623159838688065, | |
| "learning_rate": 3.95723984765825e-05, | |
| "loss": 0.1885, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17515334486961365, | |
| "step": 700, | |
| "valid_targets_mean": 4726.1, | |
| "valid_targets_min": 2134 | |
| }, | |
| { | |
| "epoch": 1.1261980830670926, | |
| "grad_norm": 0.5069059589398448, | |
| "learning_rate": 3.955585585307329e-05, | |
| "loss": 0.2061, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2065981775522232, | |
| "step": 705, | |
| "valid_targets_mean": 4385.2, | |
| "valid_targets_min": 861 | |
| }, | |
| { | |
| "epoch": 1.134185303514377, | |
| "grad_norm": 0.47661579391524955, | |
| "learning_rate": 3.9539002871915395e-05, | |
| "loss": 0.1891, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2002323418855667, | |
| "step": 710, | |
| "valid_targets_mean": 4354.8, | |
| "valid_targets_min": 998 | |
| }, | |
| { | |
| "epoch": 1.1421725239616614, | |
| "grad_norm": 0.5041034389799044, | |
| "learning_rate": 3.952183980057096e-05, | |
| "loss": 0.2032, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23501527309417725, | |
| "step": 715, | |
| "valid_targets_mean": 4711.4, | |
| "valid_targets_min": 951 | |
| }, | |
| { | |
| "epoch": 1.1501597444089458, | |
| "grad_norm": 0.4305387579316802, | |
| "learning_rate": 3.950436691142339e-05, | |
| "loss": 0.1964, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1951214224100113, | |
| "step": 720, | |
| "valid_targets_mean": 4739.7, | |
| "valid_targets_min": 2203 | |
| }, | |
| { | |
| "epoch": 1.15814696485623, | |
| "grad_norm": 0.41902185707346046, | |
| "learning_rate": 3.948658448177299e-05, | |
| "loss": 0.1885, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1951594054698944, | |
| "step": 725, | |
| "valid_targets_mean": 5117.2, | |
| "valid_targets_min": 315 | |
| }, | |
| { | |
| "epoch": 1.1661341853035143, | |
| "grad_norm": 0.4501983107651042, | |
| "learning_rate": 3.946849279383258e-05, | |
| "loss": 0.182, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1826249659061432, | |
| "step": 730, | |
| "valid_targets_mean": 5082.9, | |
| "valid_targets_min": 2936 | |
| }, | |
| { | |
| "epoch": 1.1741214057507987, | |
| "grad_norm": 0.5102934008017967, | |
| "learning_rate": 3.9450092134722984e-05, | |
| "loss": 0.1899, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20862647891044617, | |
| "step": 735, | |
| "valid_targets_mean": 4427.9, | |
| "valid_targets_min": 925 | |
| }, | |
| { | |
| "epoch": 1.182108626198083, | |
| "grad_norm": 0.37091595276571304, | |
| "learning_rate": 3.943138279646853e-05, | |
| "loss": 0.1888, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18369600176811218, | |
| "step": 740, | |
| "valid_targets_mean": 5965.6, | |
| "valid_targets_min": 2402 | |
| }, | |
| { | |
| "epoch": 1.1900958466453675, | |
| "grad_norm": 0.3828149886380875, | |
| "learning_rate": 3.941236507599234e-05, | |
| "loss": 0.1954, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1747969090938568, | |
| "step": 745, | |
| "valid_targets_mean": 5986.6, | |
| "valid_targets_min": 3205 | |
| }, | |
| { | |
| "epoch": 1.1980830670926517, | |
| "grad_norm": 0.42299079509264664, | |
| "learning_rate": 3.93930392751117e-05, | |
| "loss": 0.1919, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17547492682933807, | |
| "step": 750, | |
| "valid_targets_mean": 5405.0, | |
| "valid_targets_min": 2762 | |
| }, | |
| { | |
| "epoch": 1.206070287539936, | |
| "grad_norm": 0.5307691539881366, | |
| "learning_rate": 3.9373405700533204e-05, | |
| "loss": 0.2047, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20785072445869446, | |
| "step": 755, | |
| "valid_targets_mean": 5015.9, | |
| "valid_targets_min": 2867 | |
| }, | |
| { | |
| "epoch": 1.2140575079872205, | |
| "grad_norm": 0.45972895511875084, | |
| "learning_rate": 3.935346466384793e-05, | |
| "loss": 0.1992, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21108435094356537, | |
| "step": 760, | |
| "valid_targets_mean": 4869.1, | |
| "valid_targets_min": 1914 | |
| }, | |
| { | |
| "epoch": 1.2220447284345048, | |
| "grad_norm": 0.463583991185198, | |
| "learning_rate": 3.933321648152646e-05, | |
| "loss": 0.1888, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2199777066707611, | |
| "step": 765, | |
| "valid_targets_mean": 4760.4, | |
| "valid_targets_min": 2467 | |
| }, | |
| { | |
| "epoch": 1.230031948881789, | |
| "grad_norm": 0.5303452285663175, | |
| "learning_rate": 3.931266147491389e-05, | |
| "loss": 0.2038, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19813013076782227, | |
| "step": 770, | |
| "valid_targets_mean": 4742.1, | |
| "valid_targets_min": 1547 | |
| }, | |
| { | |
| "epoch": 1.2380191693290734, | |
| "grad_norm": 0.44210027601056695, | |
| "learning_rate": 3.929179997022471e-05, | |
| "loss": 0.1941, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19427384436130524, | |
| "step": 775, | |
| "valid_targets_mean": 4689.1, | |
| "valid_targets_min": 2415 | |
| }, | |
| { | |
| "epoch": 1.2460063897763578, | |
| "grad_norm": 0.4260036027466459, | |
| "learning_rate": 3.927063229853763e-05, | |
| "loss": 0.1888, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18853560090065002, | |
| "step": 780, | |
| "valid_targets_mean": 4814.5, | |
| "valid_targets_min": 1124 | |
| }, | |
| { | |
| "epoch": 1.2539936102236422, | |
| "grad_norm": 0.4203212640750045, | |
| "learning_rate": 3.9249158795790316e-05, | |
| "loss": 0.1893, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17745302617549896, | |
| "step": 785, | |
| "valid_targets_mean": 5653.1, | |
| "valid_targets_min": 2626 | |
| }, | |
| { | |
| "epoch": 1.2619808306709266, | |
| "grad_norm": 0.45127900485949973, | |
| "learning_rate": 3.9227379802774106e-05, | |
| "loss": 0.1873, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18783918023109436, | |
| "step": 790, | |
| "valid_targets_mean": 4700.5, | |
| "valid_targets_min": 2566 | |
| }, | |
| { | |
| "epoch": 1.269968051118211, | |
| "grad_norm": 0.4449461222766197, | |
| "learning_rate": 3.920529566512852e-05, | |
| "loss": 0.2028, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22279171645641327, | |
| "step": 795, | |
| "valid_targets_mean": 5200.9, | |
| "valid_targets_min": 2882 | |
| }, | |
| { | |
| "epoch": 1.2779552715654952, | |
| "grad_norm": 0.4427331686919491, | |
| "learning_rate": 3.918290673333585e-05, | |
| "loss": 0.1903, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21096175909042358, | |
| "step": 800, | |
| "valid_targets_mean": 5889.6, | |
| "valid_targets_min": 2399 | |
| }, | |
| { | |
| "epoch": 1.2859424920127795, | |
| "grad_norm": 0.4788651666623234, | |
| "learning_rate": 3.916021336271556e-05, | |
| "loss": 0.1924, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1883711963891983, | |
| "step": 805, | |
| "valid_targets_mean": 4681.6, | |
| "valid_targets_min": 1838 | |
| }, | |
| { | |
| "epoch": 1.293929712460064, | |
| "grad_norm": 0.42644981302965274, | |
| "learning_rate": 3.913721591341867e-05, | |
| "loss": 0.1976, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19774174690246582, | |
| "step": 810, | |
| "valid_targets_mean": 5208.6, | |
| "valid_targets_min": 882 | |
| }, | |
| { | |
| "epoch": 1.3019169329073481, | |
| "grad_norm": 0.4282394086443339, | |
| "learning_rate": 3.9113914750421985e-05, | |
| "loss": 0.1858, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2024020552635193, | |
| "step": 815, | |
| "valid_targets_mean": 5171.4, | |
| "valid_targets_min": 2203 | |
| }, | |
| { | |
| "epoch": 1.3099041533546325, | |
| "grad_norm": 0.463934425978166, | |
| "learning_rate": 3.9090310243522394e-05, | |
| "loss": 0.1916, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1852428913116455, | |
| "step": 820, | |
| "valid_targets_mean": 4476.3, | |
| "valid_targets_min": 2061 | |
| }, | |
| { | |
| "epoch": 1.317891373801917, | |
| "grad_norm": 0.4626968734289929, | |
| "learning_rate": 3.90664027673309e-05, | |
| "loss": 0.1991, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21195541322231293, | |
| "step": 825, | |
| "valid_targets_mean": 4763.8, | |
| "valid_targets_min": 328 | |
| }, | |
| { | |
| "epoch": 1.3258785942492013, | |
| "grad_norm": 0.4913124618601896, | |
| "learning_rate": 3.904219270126677e-05, | |
| "loss": 0.1831, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1818629503250122, | |
| "step": 830, | |
| "valid_targets_mean": 5537.3, | |
| "valid_targets_min": 3035 | |
| }, | |
| { | |
| "epoch": 1.3338658146964857, | |
| "grad_norm": 0.45606154339557864, | |
| "learning_rate": 3.901768042955144e-05, | |
| "loss": 0.1941, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1931591033935547, | |
| "step": 835, | |
| "valid_targets_mean": 4690.6, | |
| "valid_targets_min": 2006 | |
| }, | |
| { | |
| "epoch": 1.34185303514377, | |
| "grad_norm": 0.4206490379917601, | |
| "learning_rate": 3.8992866341202446e-05, | |
| "loss": 0.1925, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1902654767036438, | |
| "step": 840, | |
| "valid_targets_mean": 5183.4, | |
| "valid_targets_min": 1978 | |
| }, | |
| { | |
| "epoch": 1.3498402555910542, | |
| "grad_norm": 0.4100737573298077, | |
| "learning_rate": 3.8967750830027277e-05, | |
| "loss": 0.1891, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17594757676124573, | |
| "step": 845, | |
| "valid_targets_mean": 4910.4, | |
| "valid_targets_min": 2628 | |
| }, | |
| { | |
| "epoch": 1.3578274760383386, | |
| "grad_norm": 0.44034595808710914, | |
| "learning_rate": 3.894233429461706e-05, | |
| "loss": 0.1969, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18496453762054443, | |
| "step": 850, | |
| "valid_targets_mean": 4770.9, | |
| "valid_targets_min": 627 | |
| }, | |
| { | |
| "epoch": 1.365814696485623, | |
| "grad_norm": 0.44479372712144194, | |
| "learning_rate": 3.89166171383403e-05, | |
| "loss": 0.1828, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1826130747795105, | |
| "step": 855, | |
| "valid_targets_mean": 5491.2, | |
| "valid_targets_min": 1968 | |
| }, | |
| { | |
| "epoch": 1.3738019169329074, | |
| "grad_norm": 0.4251971911786779, | |
| "learning_rate": 3.889059976933644e-05, | |
| "loss": 0.2, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2241196632385254, | |
| "step": 860, | |
| "valid_targets_mean": 5514.6, | |
| "valid_targets_min": 3059 | |
| }, | |
| { | |
| "epoch": 1.3817891373801916, | |
| "grad_norm": 0.5438804780652561, | |
| "learning_rate": 3.88642826005094e-05, | |
| "loss": 0.1829, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2087685763835907, | |
| "step": 865, | |
| "valid_targets_mean": 5128.8, | |
| "valid_targets_min": 2761 | |
| }, | |
| { | |
| "epoch": 1.389776357827476, | |
| "grad_norm": 0.4536930379483279, | |
| "learning_rate": 3.883766604952102e-05, | |
| "loss": 0.1967, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17733384668827057, | |
| "step": 870, | |
| "valid_targets_mean": 5062.3, | |
| "valid_targets_min": 2239 | |
| }, | |
| { | |
| "epoch": 1.3977635782747604, | |
| "grad_norm": 0.45326614633069334, | |
| "learning_rate": 3.8810750538784404e-05, | |
| "loss": 0.1866, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18120114505290985, | |
| "step": 875, | |
| "valid_targets_mean": 4582.1, | |
| "valid_targets_min": 555 | |
| }, | |
| { | |
| "epoch": 1.4057507987220448, | |
| "grad_norm": 0.45280224587709567, | |
| "learning_rate": 3.878353649545728e-05, | |
| "loss": 0.1765, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18712976574897766, | |
| "step": 880, | |
| "valid_targets_mean": 5361.6, | |
| "valid_targets_min": 1490 | |
| }, | |
| { | |
| "epoch": 1.4137380191693292, | |
| "grad_norm": 0.3935054851636348, | |
| "learning_rate": 3.875602435143517e-05, | |
| "loss": 0.1905, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18290728330612183, | |
| "step": 885, | |
| "valid_targets_mean": 5582.3, | |
| "valid_targets_min": 2806 | |
| }, | |
| { | |
| "epoch": 1.4217252396166133, | |
| "grad_norm": 0.40395195138218626, | |
| "learning_rate": 3.872821454334453e-05, | |
| "loss": 0.1825, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1731673628091812, | |
| "step": 890, | |
| "valid_targets_mean": 5338.1, | |
| "valid_targets_min": 2288 | |
| }, | |
| { | |
| "epoch": 1.4297124600638977, | |
| "grad_norm": 0.4460038427652111, | |
| "learning_rate": 3.870010751253587e-05, | |
| "loss": 0.1938, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18078327178955078, | |
| "step": 895, | |
| "valid_targets_mean": 5166.2, | |
| "valid_targets_min": 1636 | |
| }, | |
| { | |
| "epoch": 1.4376996805111821, | |
| "grad_norm": 0.6909983676925874, | |
| "learning_rate": 3.867170370507668e-05, | |
| "loss": 0.1824, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18836957216262817, | |
| "step": 900, | |
| "valid_targets_mean": 4207.6, | |
| "valid_targets_min": 2756 | |
| }, | |
| { | |
| "epoch": 1.4456869009584665, | |
| "grad_norm": 0.4890825578133814, | |
| "learning_rate": 3.8643003571744445e-05, | |
| "loss": 0.202, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20180103182792664, | |
| "step": 905, | |
| "valid_targets_mean": 5083.1, | |
| "valid_targets_min": 2853 | |
| }, | |
| { | |
| "epoch": 1.4536741214057507, | |
| "grad_norm": 0.44165125354651835, | |
| "learning_rate": 3.861400756801938e-05, | |
| "loss": 0.1971, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18903714418411255, | |
| "step": 910, | |
| "valid_targets_mean": 5027.4, | |
| "valid_targets_min": 2207 | |
| }, | |
| { | |
| "epoch": 1.461661341853035, | |
| "grad_norm": 0.42365517800692115, | |
| "learning_rate": 3.8584716154077286e-05, | |
| "loss": 0.1994, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18654455244541168, | |
| "step": 915, | |
| "valid_targets_mean": 4939.5, | |
| "valid_targets_min": 1032 | |
| }, | |
| { | |
| "epoch": 1.4696485623003195, | |
| "grad_norm": 0.4064283909453683, | |
| "learning_rate": 3.855512979478222e-05, | |
| "loss": 0.1908, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18689365684986115, | |
| "step": 920, | |
| "valid_targets_mean": 5333.6, | |
| "valid_targets_min": 2296 | |
| }, | |
| { | |
| "epoch": 1.4776357827476039, | |
| "grad_norm": 0.5171283346007386, | |
| "learning_rate": 3.852524895967911e-05, | |
| "loss": 0.2003, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1942915916442871, | |
| "step": 925, | |
| "valid_targets_mean": 5256.5, | |
| "valid_targets_min": 1538 | |
| }, | |
| { | |
| "epoch": 1.4856230031948883, | |
| "grad_norm": 0.4346617570847732, | |
| "learning_rate": 3.8495074122986296e-05, | |
| "loss": 0.1811, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18241260945796967, | |
| "step": 930, | |
| "valid_targets_mean": 4790.5, | |
| "valid_targets_min": 2794 | |
| }, | |
| { | |
| "epoch": 1.4936102236421724, | |
| "grad_norm": 0.4305348916345901, | |
| "learning_rate": 3.846460576358804e-05, | |
| "loss": 0.1913, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20855984091758728, | |
| "step": 935, | |
| "valid_targets_mean": 5686.1, | |
| "valid_targets_min": 2197 | |
| }, | |
| { | |
| "epoch": 1.5015974440894568, | |
| "grad_norm": 0.4254847491946138, | |
| "learning_rate": 3.843384436502688e-05, | |
| "loss": 0.2057, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18769283592700958, | |
| "step": 940, | |
| "valid_targets_mean": 5134.1, | |
| "valid_targets_min": 1776 | |
| }, | |
| { | |
| "epoch": 1.5095846645367412, | |
| "grad_norm": 0.4959658900898383, | |
| "learning_rate": 3.8402790415496e-05, | |
| "loss": 0.1939, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25189208984375, | |
| "step": 945, | |
| "valid_targets_mean": 4566.4, | |
| "valid_targets_min": 2620 | |
| }, | |
| { | |
| "epoch": 1.5175718849840254, | |
| "grad_norm": 0.45091193602879176, | |
| "learning_rate": 3.837144440783144e-05, | |
| "loss": 0.1874, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18159735202789307, | |
| "step": 950, | |
| "valid_targets_mean": 5124.5, | |
| "valid_targets_min": 2501 | |
| }, | |
| { | |
| "epoch": 1.5255591054313098, | |
| "grad_norm": 0.4455240816070557, | |
| "learning_rate": 3.833980683950431e-05, | |
| "loss": 0.1914, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18155726790428162, | |
| "step": 955, | |
| "valid_targets_mean": 4296.4, | |
| "valid_targets_min": 1278 | |
| }, | |
| { | |
| "epoch": 1.5335463258785942, | |
| "grad_norm": 0.4967898558869729, | |
| "learning_rate": 3.8307878212612886e-05, | |
| "loss": 0.1822, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18451157212257385, | |
| "step": 960, | |
| "valid_targets_mean": 5424.6, | |
| "valid_targets_min": 1815 | |
| }, | |
| { | |
| "epoch": 1.5415335463258786, | |
| "grad_norm": 0.5705405559639737, | |
| "learning_rate": 3.827565903387461e-05, | |
| "loss": 0.1743, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17551621794700623, | |
| "step": 965, | |
| "valid_targets_mean": 5158.1, | |
| "valid_targets_min": 1909 | |
| }, | |
| { | |
| "epoch": 1.549520766773163, | |
| "grad_norm": 0.439193655777034, | |
| "learning_rate": 3.82431498146181e-05, | |
| "loss": 0.1806, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17131733894348145, | |
| "step": 970, | |
| "valid_targets_mean": 5127.6, | |
| "valid_targets_min": 3056 | |
| }, | |
| { | |
| "epoch": 1.5575079872204474, | |
| "grad_norm": 0.3866868535101479, | |
| "learning_rate": 3.821035107077499e-05, | |
| "loss": 0.1853, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17706656455993652, | |
| "step": 975, | |
| "valid_targets_mean": 5056.6, | |
| "valid_targets_min": 918 | |
| }, | |
| { | |
| "epoch": 1.5654952076677318, | |
| "grad_norm": 0.44096498809755014, | |
| "learning_rate": 3.817726332287179e-05, | |
| "loss": 0.1799, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19993554055690765, | |
| "step": 980, | |
| "valid_targets_mean": 5096.1, | |
| "valid_targets_min": 723 | |
| }, | |
| { | |
| "epoch": 1.573482428115016, | |
| "grad_norm": 0.39177645906885344, | |
| "learning_rate": 3.814388709602155e-05, | |
| "loss": 0.1746, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14458595216274261, | |
| "step": 985, | |
| "valid_targets_mean": 4329.7, | |
| "valid_targets_min": 620 | |
| }, | |
| { | |
| "epoch": 1.5814696485623003, | |
| "grad_norm": 0.3940026421236249, | |
| "learning_rate": 3.8110222919915626e-05, | |
| "loss": 0.1868, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17335283756256104, | |
| "step": 990, | |
| "valid_targets_mean": 5078.6, | |
| "valid_targets_min": 2601 | |
| }, | |
| { | |
| "epoch": 1.5894568690095847, | |
| "grad_norm": 0.4012383401813614, | |
| "learning_rate": 3.8076271328815175e-05, | |
| "loss": 0.1901, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19600173830986023, | |
| "step": 995, | |
| "valid_targets_mean": 5560.4, | |
| "valid_targets_min": 3864 | |
| }, | |
| { | |
| "epoch": 1.5974440894568689, | |
| "grad_norm": 0.4452269513403438, | |
| "learning_rate": 3.804203286154275e-05, | |
| "loss": 0.1893, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20371435582637787, | |
| "step": 1000, | |
| "valid_targets_mean": 5266.3, | |
| "valid_targets_min": 2530 | |
| }, | |
| { | |
| "epoch": 1.6054313099041533, | |
| "grad_norm": 0.40601362352375825, | |
| "learning_rate": 3.800750806147371e-05, | |
| "loss": 0.1998, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17152462899684906, | |
| "step": 1005, | |
| "valid_targets_mean": 5045.0, | |
| "valid_targets_min": 2955 | |
| }, | |
| { | |
| "epoch": 1.6134185303514377, | |
| "grad_norm": 0.7843507911699634, | |
| "learning_rate": 3.79726974765276e-05, | |
| "loss": 0.1794, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1938866674900055, | |
| "step": 1010, | |
| "valid_targets_mean": 4996.1, | |
| "valid_targets_min": 969 | |
| }, | |
| { | |
| "epoch": 1.621405750798722, | |
| "grad_norm": 0.40847946755087117, | |
| "learning_rate": 3.793760165915947e-05, | |
| "loss": 0.1871, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1732960343360901, | |
| "step": 1015, | |
| "valid_targets_mean": 4813.6, | |
| "valid_targets_min": 2006 | |
| }, | |
| { | |
| "epoch": 1.6293929712460065, | |
| "grad_norm": 0.3918308223232367, | |
| "learning_rate": 3.7902221166351106e-05, | |
| "loss": 0.1894, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21553091704845428, | |
| "step": 1020, | |
| "valid_targets_mean": 5718.3, | |
| "valid_targets_min": 3055 | |
| }, | |
| { | |
| "epoch": 1.6373801916932909, | |
| "grad_norm": 0.411913328655, | |
| "learning_rate": 3.786655655960216e-05, | |
| "loss": 0.1802, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20339325070381165, | |
| "step": 1025, | |
| "valid_targets_mean": 5635.0, | |
| "valid_targets_min": 2501 | |
| }, | |
| { | |
| "epoch": 1.645367412140575, | |
| "grad_norm": 0.39626617317609736, | |
| "learning_rate": 3.7830608404921294e-05, | |
| "loss": 0.2028, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21556724607944489, | |
| "step": 1030, | |
| "valid_targets_mean": 5670.8, | |
| "valid_targets_min": 1412 | |
| }, | |
| { | |
| "epoch": 1.6533546325878594, | |
| "grad_norm": 0.4012601564655484, | |
| "learning_rate": 3.7794377272817144e-05, | |
| "loss": 0.1818, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19218207895755768, | |
| "step": 1035, | |
| "valid_targets_mean": 5394.6, | |
| "valid_targets_min": 2899 | |
| }, | |
| { | |
| "epoch": 1.6613418530351438, | |
| "grad_norm": 0.3940816896663035, | |
| "learning_rate": 3.775786373828929e-05, | |
| "loss": 0.1784, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18354302644729614, | |
| "step": 1040, | |
| "valid_targets_mean": 5949.4, | |
| "valid_targets_min": 2658 | |
| }, | |
| { | |
| "epoch": 1.669329073482428, | |
| "grad_norm": 0.3769212874880954, | |
| "learning_rate": 3.7721068380819135e-05, | |
| "loss": 0.1751, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16321788728237152, | |
| "step": 1045, | |
| "valid_targets_mean": 5054.8, | |
| "valid_targets_min": 2060 | |
| }, | |
| { | |
| "epoch": 1.6773162939297124, | |
| "grad_norm": 0.40117242238353595, | |
| "learning_rate": 3.768399178436069e-05, | |
| "loss": 0.1761, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1871631145477295, | |
| "step": 1050, | |
| "valid_targets_mean": 4880.2, | |
| "valid_targets_min": 326 | |
| }, | |
| { | |
| "epoch": 1.6853035143769968, | |
| "grad_norm": 0.3551203405093787, | |
| "learning_rate": 3.764663453733135e-05, | |
| "loss": 0.1752, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16526569426059723, | |
| "step": 1055, | |
| "valid_targets_mean": 5719.8, | |
| "valid_targets_min": 2671 | |
| }, | |
| { | |
| "epoch": 1.6932907348242812, | |
| "grad_norm": 0.41226459204991145, | |
| "learning_rate": 3.7608997232602475e-05, | |
| "loss": 0.2039, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18004834651947021, | |
| "step": 1060, | |
| "valid_targets_mean": 5151.2, | |
| "valid_targets_min": 1771 | |
| }, | |
| { | |
| "epoch": 1.7012779552715656, | |
| "grad_norm": 0.4264168040457079, | |
| "learning_rate": 3.757108046749006e-05, | |
| "loss": 0.1842, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18524786829948425, | |
| "step": 1065, | |
| "valid_targets_mean": 4731.6, | |
| "valid_targets_min": 2804 | |
| }, | |
| { | |
| "epoch": 1.70926517571885, | |
| "grad_norm": 0.43194953924536067, | |
| "learning_rate": 3.753288484374524e-05, | |
| "loss": 0.1968, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19120649993419647, | |
| "step": 1070, | |
| "valid_targets_mean": 4892.5, | |
| "valid_targets_min": 1374 | |
| }, | |
| { | |
| "epoch": 1.7172523961661343, | |
| "grad_norm": 0.4576001359550158, | |
| "learning_rate": 3.7494410967544674e-05, | |
| "loss": 0.1766, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16587048768997192, | |
| "step": 1075, | |
| "valid_targets_mean": 4434.6, | |
| "valid_targets_min": 348 | |
| }, | |
| { | |
| "epoch": 1.7252396166134185, | |
| "grad_norm": 0.49177275071095927, | |
| "learning_rate": 3.745565944948103e-05, | |
| "loss": 0.1956, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1851913183927536, | |
| "step": 1080, | |
| "valid_targets_mean": 4221.4, | |
| "valid_targets_min": 648 | |
| }, | |
| { | |
| "epoch": 1.733226837060703, | |
| "grad_norm": 0.45490629524482995, | |
| "learning_rate": 3.7416630904553205e-05, | |
| "loss": 0.1893, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20662656426429749, | |
| "step": 1085, | |
| "valid_targets_mean": 4479.7, | |
| "valid_targets_min": 1065 | |
| }, | |
| { | |
| "epoch": 1.741214057507987, | |
| "grad_norm": 0.4850957700891692, | |
| "learning_rate": 3.737732595215663e-05, | |
| "loss": 0.1781, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20295533537864685, | |
| "step": 1090, | |
| "valid_targets_mean": 4505.8, | |
| "valid_targets_min": 2520 | |
| }, | |
| { | |
| "epoch": 1.7492012779552715, | |
| "grad_norm": 0.38641912918615196, | |
| "learning_rate": 3.733774521607338e-05, | |
| "loss": 0.183, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1817728877067566, | |
| "step": 1095, | |
| "valid_targets_mean": 5386.7, | |
| "valid_targets_min": 1158 | |
| }, | |
| { | |
| "epoch": 1.7571884984025559, | |
| "grad_norm": 0.3936030431480062, | |
| "learning_rate": 3.729788932446231e-05, | |
| "loss": 0.177, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16919134557247162, | |
| "step": 1100, | |
| "valid_targets_mean": 5214.6, | |
| "valid_targets_min": 2954 | |
| }, | |
| { | |
| "epoch": 1.7651757188498403, | |
| "grad_norm": 0.5153068926934984, | |
| "learning_rate": 3.7257758909849107e-05, | |
| "loss": 0.1807, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17329710721969604, | |
| "step": 1105, | |
| "valid_targets_mean": 5111.8, | |
| "valid_targets_min": 1093 | |
| }, | |
| { | |
| "epoch": 1.7731629392971247, | |
| "grad_norm": 0.394871232433515, | |
| "learning_rate": 3.72173546091162e-05, | |
| "loss": 0.1814, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16914108395576477, | |
| "step": 1110, | |
| "valid_targets_mean": 4976.5, | |
| "valid_targets_min": 2562 | |
| }, | |
| { | |
| "epoch": 1.781150159744409, | |
| "grad_norm": 0.38395016528571063, | |
| "learning_rate": 3.7176677063492685e-05, | |
| "loss": 0.1776, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16938599944114685, | |
| "step": 1115, | |
| "valid_targets_mean": 4870.1, | |
| "valid_targets_min": 2089 | |
| }, | |
| { | |
| "epoch": 1.7891373801916934, | |
| "grad_norm": 0.4660291326195513, | |
| "learning_rate": 3.713572691854414e-05, | |
| "loss": 0.187, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1740354299545288, | |
| "step": 1120, | |
| "valid_targets_mean": 4758.6, | |
| "valid_targets_min": 2011 | |
| }, | |
| { | |
| "epoch": 1.7971246006389776, | |
| "grad_norm": 0.4760954631530669, | |
| "learning_rate": 3.709450482416239e-05, | |
| "loss": 0.1797, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16924232244491577, | |
| "step": 1125, | |
| "valid_targets_mean": 4782.5, | |
| "valid_targets_min": 236 | |
| }, | |
| { | |
| "epoch": 1.805111821086262, | |
| "grad_norm": 0.44953831601130506, | |
| "learning_rate": 3.7053011434555165e-05, | |
| "loss": 0.1914, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1874721199274063, | |
| "step": 1130, | |
| "valid_targets_mean": 4750.6, | |
| "valid_targets_min": 1877 | |
| }, | |
| { | |
| "epoch": 1.8130990415335462, | |
| "grad_norm": 0.4237752285270204, | |
| "learning_rate": 3.701124740823575e-05, | |
| "loss": 0.189, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19240900874137878, | |
| "step": 1135, | |
| "valid_targets_mean": 4912.1, | |
| "valid_targets_min": 2797 | |
| }, | |
| { | |
| "epoch": 1.8210862619808306, | |
| "grad_norm": 0.39974962472469494, | |
| "learning_rate": 3.696921340801253e-05, | |
| "loss": 0.1875, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1687966287136078, | |
| "step": 1140, | |
| "valid_targets_mean": 5087.6, | |
| "valid_targets_min": 2253 | |
| }, | |
| { | |
| "epoch": 1.829073482428115, | |
| "grad_norm": 0.46128284504636435, | |
| "learning_rate": 3.6926910100978444e-05, | |
| "loss": 0.1747, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1693594604730606, | |
| "step": 1145, | |
| "valid_targets_mean": 4924.9, | |
| "valid_targets_min": 2138 | |
| }, | |
| { | |
| "epoch": 1.8370607028753994, | |
| "grad_norm": 0.3966679891925233, | |
| "learning_rate": 3.688433815850041e-05, | |
| "loss": 0.195, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17254772782325745, | |
| "step": 1150, | |
| "valid_targets_mean": 5032.2, | |
| "valid_targets_min": 2388 | |
| }, | |
| { | |
| "epoch": 1.8450479233226837, | |
| "grad_norm": 0.404262990463488, | |
| "learning_rate": 3.68414982562087e-05, | |
| "loss": 0.1814, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17766813933849335, | |
| "step": 1155, | |
| "valid_targets_mean": 4867.5, | |
| "valid_targets_min": 3482 | |
| }, | |
| { | |
| "epoch": 1.8530351437699681, | |
| "grad_norm": 0.4662877485740305, | |
| "learning_rate": 3.679839107398618e-05, | |
| "loss": 0.1823, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18041929602622986, | |
| "step": 1160, | |
| "valid_targets_mean": 4748.1, | |
| "valid_targets_min": 1879 | |
| }, | |
| { | |
| "epoch": 1.8610223642172525, | |
| "grad_norm": 0.3990535545472509, | |
| "learning_rate": 3.6755017295957536e-05, | |
| "loss": 0.1745, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16708019375801086, | |
| "step": 1165, | |
| "valid_targets_mean": 4892.9, | |
| "valid_targets_min": 2470 | |
| }, | |
| { | |
| "epoch": 1.8690095846645367, | |
| "grad_norm": 0.4144209385591321, | |
| "learning_rate": 3.67113776104784e-05, | |
| "loss": 0.1916, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17229919135570526, | |
| "step": 1170, | |
| "valid_targets_mean": 5163.4, | |
| "valid_targets_min": 2455 | |
| }, | |
| { | |
| "epoch": 1.876996805111821, | |
| "grad_norm": 0.46583395392864874, | |
| "learning_rate": 3.6667472710124475e-05, | |
| "loss": 0.196, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22971127927303314, | |
| "step": 1175, | |
| "valid_targets_mean": 4816.1, | |
| "valid_targets_min": 736 | |
| }, | |
| { | |
| "epoch": 1.8849840255591053, | |
| "grad_norm": 0.38387303239642573, | |
| "learning_rate": 3.6623303291680465e-05, | |
| "loss": 0.1812, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1724814474582672, | |
| "step": 1180, | |
| "valid_targets_mean": 5413.2, | |
| "valid_targets_min": 1916 | |
| }, | |
| { | |
| "epoch": 1.8929712460063897, | |
| "grad_norm": 0.4238761492264338, | |
| "learning_rate": 3.6578870056129086e-05, | |
| "loss": 0.1886, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20026056468486786, | |
| "step": 1185, | |
| "valid_targets_mean": 4938.4, | |
| "valid_targets_min": 903 | |
| }, | |
| { | |
| "epoch": 1.900958466453674, | |
| "grad_norm": 0.3856692370786231, | |
| "learning_rate": 3.653417370863992e-05, | |
| "loss": 0.1695, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1655639111995697, | |
| "step": 1190, | |
| "valid_targets_mean": 5610.5, | |
| "valid_targets_min": 2455 | |
| }, | |
| { | |
| "epoch": 1.9089456869009584, | |
| "grad_norm": 0.43703245766439625, | |
| "learning_rate": 3.6489214958558206e-05, | |
| "loss": 0.202, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18975260853767395, | |
| "step": 1195, | |
| "valid_targets_mean": 4508.4, | |
| "valid_targets_min": 1745 | |
| }, | |
| { | |
| "epoch": 1.9169329073482428, | |
| "grad_norm": 0.5281049511018331, | |
| "learning_rate": 3.644399451939358e-05, | |
| "loss": 0.1887, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21455372869968414, | |
| "step": 1200, | |
| "valid_targets_mean": 4956.7, | |
| "valid_targets_min": 801 | |
| }, | |
| { | |
| "epoch": 1.9249201277955272, | |
| "grad_norm": 0.4456116048169732, | |
| "learning_rate": 3.639851310880881e-05, | |
| "loss": 0.1989, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22581738233566284, | |
| "step": 1205, | |
| "valid_targets_mean": 5078.5, | |
| "valid_targets_min": 2705 | |
| }, | |
| { | |
| "epoch": 1.9329073482428116, | |
| "grad_norm": 0.44347445744978314, | |
| "learning_rate": 3.635277144860834e-05, | |
| "loss": 0.1729, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18027453124523163, | |
| "step": 1210, | |
| "valid_targets_mean": 5286.9, | |
| "valid_targets_min": 3672 | |
| }, | |
| { | |
| "epoch": 1.9408945686900958, | |
| "grad_norm": 0.45736753248946777, | |
| "learning_rate": 3.630677026472684e-05, | |
| "loss": 0.1968, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18593823909759521, | |
| "step": 1215, | |
| "valid_targets_mean": 4954.5, | |
| "valid_targets_min": 1976 | |
| }, | |
| { | |
| "epoch": 1.9488817891373802, | |
| "grad_norm": 0.44589588400860614, | |
| "learning_rate": 3.626051028721773e-05, | |
| "loss": 0.1854, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17879164218902588, | |
| "step": 1220, | |
| "valid_targets_mean": 4284.3, | |
| "valid_targets_min": 991 | |
| }, | |
| { | |
| "epoch": 1.9568690095846646, | |
| "grad_norm": 0.4066165331003282, | |
| "learning_rate": 3.621399225024156e-05, | |
| "loss": 0.1938, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21019574999809265, | |
| "step": 1225, | |
| "valid_targets_mean": 5225.5, | |
| "valid_targets_min": 2291 | |
| }, | |
| { | |
| "epoch": 1.9648562300319488, | |
| "grad_norm": 0.45244007382089657, | |
| "learning_rate": 3.616721689205436e-05, | |
| "loss": 0.1643, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15397858619689941, | |
| "step": 1230, | |
| "valid_targets_mean": 4991.4, | |
| "valid_targets_min": 2476 | |
| }, | |
| { | |
| "epoch": 1.9728434504792332, | |
| "grad_norm": 0.41017598135316413, | |
| "learning_rate": 3.612018495499594e-05, | |
| "loss": 0.1731, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17426639795303345, | |
| "step": 1235, | |
| "valid_targets_mean": 5335.1, | |
| "valid_targets_min": 1954 | |
| }, | |
| { | |
| "epoch": 1.9808306709265175, | |
| "grad_norm": 0.39867931036265597, | |
| "learning_rate": 3.6072897185478074e-05, | |
| "loss": 0.1831, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17372800409793854, | |
| "step": 1240, | |
| "valid_targets_mean": 5125.2, | |
| "valid_targets_min": 2162 | |
| }, | |
| { | |
| "epoch": 1.988817891373802, | |
| "grad_norm": 0.395402496310432, | |
| "learning_rate": 3.6025354333972714e-05, | |
| "loss": 0.1866, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20539633929729462, | |
| "step": 1245, | |
| "valid_targets_mean": 5286.0, | |
| "valid_targets_min": 2232 | |
| }, | |
| { | |
| "epoch": 1.9968051118210863, | |
| "grad_norm": 0.4217228255263356, | |
| "learning_rate": 3.597755715500002e-05, | |
| "loss": 0.1774, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19640657305717468, | |
| "step": 1250, | |
| "valid_targets_mean": 5069.9, | |
| "valid_targets_min": 3128 | |
| }, | |
| { | |
| "epoch": 2.0047923322683707, | |
| "grad_norm": 0.3853703821280624, | |
| "learning_rate": 3.592950640711642e-05, | |
| "loss": 0.1625, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15772667527198792, | |
| "step": 1255, | |
| "valid_targets_mean": 5379.5, | |
| "valid_targets_min": 1927 | |
| }, | |
| { | |
| "epoch": 2.012779552715655, | |
| "grad_norm": 0.5160456901623796, | |
| "learning_rate": 3.5881202852902543e-05, | |
| "loss": 0.1613, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15113511681556702, | |
| "step": 1260, | |
| "valid_targets_mean": 4432.5, | |
| "valid_targets_min": 1539 | |
| }, | |
| { | |
| "epoch": 2.0207667731629395, | |
| "grad_norm": 0.43590478944501126, | |
| "learning_rate": 3.583264725895117e-05, | |
| "loss": 0.1686, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1535935252904892, | |
| "step": 1265, | |
| "valid_targets_mean": 5177.8, | |
| "valid_targets_min": 2515 | |
| }, | |
| { | |
| "epoch": 2.0287539936102235, | |
| "grad_norm": 0.407257198892713, | |
| "learning_rate": 3.5783840395854984e-05, | |
| "loss": 0.1687, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18131369352340698, | |
| "step": 1270, | |
| "valid_targets_mean": 5044.1, | |
| "valid_targets_min": 2654 | |
| }, | |
| { | |
| "epoch": 2.036741214057508, | |
| "grad_norm": 0.41333776318949583, | |
| "learning_rate": 3.5734783038194446e-05, | |
| "loss": 0.1828, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16500158607959747, | |
| "step": 1275, | |
| "valid_targets_mean": 5371.2, | |
| "valid_targets_min": 2354 | |
| }, | |
| { | |
| "epoch": 2.0447284345047922, | |
| "grad_norm": 0.5066992622702711, | |
| "learning_rate": 3.5685475964525396e-05, | |
| "loss": 0.1589, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17958015203475952, | |
| "step": 1280, | |
| "valid_targets_mean": 5096.2, | |
| "valid_targets_min": 2031 | |
| }, | |
| { | |
| "epoch": 2.0527156549520766, | |
| "grad_norm": 0.41525107559130636, | |
| "learning_rate": 3.5635919957366775e-05, | |
| "loss": 0.1737, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16494354605674744, | |
| "step": 1285, | |
| "valid_targets_mean": 5180.6, | |
| "valid_targets_min": 1735 | |
| }, | |
| { | |
| "epoch": 2.060702875399361, | |
| "grad_norm": 0.4030564745555338, | |
| "learning_rate": 3.558611580318817e-05, | |
| "loss": 0.1766, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16279885172843933, | |
| "step": 1290, | |
| "valid_targets_mean": 5654.5, | |
| "valid_targets_min": 399 | |
| }, | |
| { | |
| "epoch": 2.0686900958466454, | |
| "grad_norm": 0.38291695695184597, | |
| "learning_rate": 3.5536064292397334e-05, | |
| "loss": 0.1707, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15881040692329407, | |
| "step": 1295, | |
| "valid_targets_mean": 5240.3, | |
| "valid_targets_min": 1571 | |
| }, | |
| { | |
| "epoch": 2.07667731629393, | |
| "grad_norm": 0.5184078869284591, | |
| "learning_rate": 3.548576621932766e-05, | |
| "loss": 0.1582, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14718276262283325, | |
| "step": 1300, | |
| "valid_targets_mean": 5209.1, | |
| "valid_targets_min": 1996 | |
| }, | |
| { | |
| "epoch": 2.084664536741214, | |
| "grad_norm": 0.4304703422247274, | |
| "learning_rate": 3.543522238222555e-05, | |
| "loss": 0.1729, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17051589488983154, | |
| "step": 1305, | |
| "valid_targets_mean": 4979.6, | |
| "valid_targets_min": 2205 | |
| }, | |
| { | |
| "epoch": 2.0926517571884986, | |
| "grad_norm": 0.509154087616053, | |
| "learning_rate": 3.538443358323777e-05, | |
| "loss": 0.1616, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15125590562820435, | |
| "step": 1310, | |
| "valid_targets_mean": 4807.1, | |
| "valid_targets_min": 2634 | |
| }, | |
| { | |
| "epoch": 2.1006389776357826, | |
| "grad_norm": 0.4027297232435667, | |
| "learning_rate": 3.5333400628398696e-05, | |
| "loss": 0.1712, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14137303829193115, | |
| "step": 1315, | |
| "valid_targets_mean": 5286.8, | |
| "valid_targets_min": 2016 | |
| }, | |
| { | |
| "epoch": 2.108626198083067, | |
| "grad_norm": 0.4299348886566083, | |
| "learning_rate": 3.5282124327617556e-05, | |
| "loss": 0.1718, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18772245943546295, | |
| "step": 1320, | |
| "valid_targets_mean": 5604.6, | |
| "valid_targets_min": 2203 | |
| }, | |
| { | |
| "epoch": 2.1166134185303513, | |
| "grad_norm": 0.3692162672370612, | |
| "learning_rate": 3.5230605494665535e-05, | |
| "loss": 0.1554, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15149496495723724, | |
| "step": 1325, | |
| "valid_targets_mean": 5460.3, | |
| "valid_targets_min": 2240 | |
| }, | |
| { | |
| "epoch": 2.1246006389776357, | |
| "grad_norm": 0.4087624988386758, | |
| "learning_rate": 3.517884494716289e-05, | |
| "loss": 0.1688, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14947164058685303, | |
| "step": 1330, | |
| "valid_targets_mean": 5299.7, | |
| "valid_targets_min": 882 | |
| }, | |
| { | |
| "epoch": 2.13258785942492, | |
| "grad_norm": 0.4189754122982718, | |
| "learning_rate": 3.512684350656596e-05, | |
| "loss": 0.1775, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17300669848918915, | |
| "step": 1335, | |
| "valid_targets_mean": 5059.1, | |
| "valid_targets_min": 374 | |
| }, | |
| { | |
| "epoch": 2.1405750798722045, | |
| "grad_norm": 0.4588574317674406, | |
| "learning_rate": 3.507460199815414e-05, | |
| "loss": 0.173, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1862129271030426, | |
| "step": 1340, | |
| "valid_targets_mean": 4647.7, | |
| "valid_targets_min": 867 | |
| }, | |
| { | |
| "epoch": 2.148562300319489, | |
| "grad_norm": 0.39974226360290294, | |
| "learning_rate": 3.5022121251016786e-05, | |
| "loss": 0.1668, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17242909967899323, | |
| "step": 1345, | |
| "valid_targets_mean": 5625.0, | |
| "valid_targets_min": 2914 | |
| }, | |
| { | |
| "epoch": 2.1565495207667733, | |
| "grad_norm": 0.43563984699739644, | |
| "learning_rate": 3.4969402098040015e-05, | |
| "loss": 0.1564, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14147107303142548, | |
| "step": 1350, | |
| "valid_targets_mean": 6062.9, | |
| "valid_targets_min": 2678 | |
| }, | |
| { | |
| "epoch": 2.1645367412140577, | |
| "grad_norm": 0.4347710448758343, | |
| "learning_rate": 3.491644537589356e-05, | |
| "loss": 0.1769, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16159726679325104, | |
| "step": 1355, | |
| "valid_targets_mean": 5487.6, | |
| "valid_targets_min": 2391 | |
| }, | |
| { | |
| "epoch": 2.1725239616613417, | |
| "grad_norm": 0.40251293787305714, | |
| "learning_rate": 3.486325192501743e-05, | |
| "loss": 0.1662, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14349807798862457, | |
| "step": 1360, | |
| "valid_targets_mean": 5164.7, | |
| "valid_targets_min": 1895 | |
| }, | |
| { | |
| "epoch": 2.180511182108626, | |
| "grad_norm": 0.4052169811831424, | |
| "learning_rate": 3.480982258960859e-05, | |
| "loss": 0.1578, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1655794084072113, | |
| "step": 1365, | |
| "valid_targets_mean": 5415.4, | |
| "valid_targets_min": 3295 | |
| }, | |
| { | |
| "epoch": 2.1884984025559104, | |
| "grad_norm": 0.3868516415857838, | |
| "learning_rate": 3.47561582176076e-05, | |
| "loss": 0.1665, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16756269335746765, | |
| "step": 1370, | |
| "valid_targets_mean": 5528.4, | |
| "valid_targets_min": 1276 | |
| }, | |
| { | |
| "epoch": 2.196485623003195, | |
| "grad_norm": 0.38658950090113675, | |
| "learning_rate": 3.470225966068507e-05, | |
| "loss": 0.1637, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15978050231933594, | |
| "step": 1375, | |
| "valid_targets_mean": 5881.1, | |
| "valid_targets_min": 1407 | |
| }, | |
| { | |
| "epoch": 2.2044728434504792, | |
| "grad_norm": 0.41106469677049956, | |
| "learning_rate": 3.464812777422826e-05, | |
| "loss": 0.1661, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16382162272930145, | |
| "step": 1380, | |
| "valid_targets_mean": 5139.6, | |
| "valid_targets_min": 1879 | |
| }, | |
| { | |
| "epoch": 2.2124600638977636, | |
| "grad_norm": 0.4409263969650839, | |
| "learning_rate": 3.459376341732741e-05, | |
| "loss": 0.1709, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17837923765182495, | |
| "step": 1385, | |
| "valid_targets_mean": 5087.9, | |
| "valid_targets_min": 2653 | |
| }, | |
| { | |
| "epoch": 2.220447284345048, | |
| "grad_norm": 0.4037924947719911, | |
| "learning_rate": 3.453916745276213e-05, | |
| "loss": 0.1708, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16253410279750824, | |
| "step": 1390, | |
| "valid_targets_mean": 4782.2, | |
| "valid_targets_min": 2050 | |
| }, | |
| { | |
| "epoch": 2.2284345047923324, | |
| "grad_norm": 0.4234071669100518, | |
| "learning_rate": 3.4484340746987745e-05, | |
| "loss": 0.1867, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16547849774360657, | |
| "step": 1395, | |
| "valid_targets_mean": 5021.2, | |
| "valid_targets_min": 1783 | |
| }, | |
| { | |
| "epoch": 2.236421725239617, | |
| "grad_norm": 0.3893125972333106, | |
| "learning_rate": 3.4429284170121494e-05, | |
| "loss": 0.175, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15550661087036133, | |
| "step": 1400, | |
| "valid_targets_mean": 5635.1, | |
| "valid_targets_min": 3904 | |
| }, | |
| { | |
| "epoch": 2.244408945686901, | |
| "grad_norm": 0.43839087800918086, | |
| "learning_rate": 3.4373998595928764e-05, | |
| "loss": 0.1802, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20556634664535522, | |
| "step": 1405, | |
| "valid_targets_mean": 4605.0, | |
| "valid_targets_min": 2835 | |
| }, | |
| { | |
| "epoch": 2.252396166134185, | |
| "grad_norm": 0.43378509816862304, | |
| "learning_rate": 3.431848490180919e-05, | |
| "loss": 0.1699, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17661327123641968, | |
| "step": 1410, | |
| "valid_targets_mean": 4568.9, | |
| "valid_targets_min": 416 | |
| }, | |
| { | |
| "epoch": 2.2603833865814695, | |
| "grad_norm": 0.4223084977697296, | |
| "learning_rate": 3.4262743968782736e-05, | |
| "loss": 0.1785, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18202045559883118, | |
| "step": 1415, | |
| "valid_targets_mean": 4809.2, | |
| "valid_targets_min": 1600 | |
| }, | |
| { | |
| "epoch": 2.268370607028754, | |
| "grad_norm": 0.40042104427089775, | |
| "learning_rate": 3.4206776681475737e-05, | |
| "loss": 0.1666, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15740567445755005, | |
| "step": 1420, | |
| "valid_targets_mean": 5083.4, | |
| "valid_targets_min": 1860 | |
| }, | |
| { | |
| "epoch": 2.2763578274760383, | |
| "grad_norm": 0.3948770061121089, | |
| "learning_rate": 3.415058392810682e-05, | |
| "loss": 0.1706, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16975916922092438, | |
| "step": 1425, | |
| "valid_targets_mean": 5008.1, | |
| "valid_targets_min": 1989 | |
| }, | |
| { | |
| "epoch": 2.2843450479233227, | |
| "grad_norm": 0.6246912923551692, | |
| "learning_rate": 3.409416660047286e-05, | |
| "loss": 0.1722, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18752746284008026, | |
| "step": 1430, | |
| "valid_targets_mean": 4607.5, | |
| "valid_targets_min": 2134 | |
| }, | |
| { | |
| "epoch": 2.292332268370607, | |
| "grad_norm": 0.4405119588607242, | |
| "learning_rate": 3.403752559393477e-05, | |
| "loss": 0.1746, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18587929010391235, | |
| "step": 1435, | |
| "valid_targets_mean": 4850.5, | |
| "valid_targets_min": 2319 | |
| }, | |
| { | |
| "epoch": 2.3003194888178915, | |
| "grad_norm": 0.4861428404104646, | |
| "learning_rate": 3.398066180740332e-05, | |
| "loss": 0.1785, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2109757959842682, | |
| "step": 1440, | |
| "valid_targets_mean": 4607.2, | |
| "valid_targets_min": 326 | |
| }, | |
| { | |
| "epoch": 2.308306709265176, | |
| "grad_norm": 0.4264456141646522, | |
| "learning_rate": 3.39235761433249e-05, | |
| "loss": 0.1677, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17636804282665253, | |
| "step": 1445, | |
| "valid_targets_mean": 4809.9, | |
| "valid_targets_min": 2170 | |
| }, | |
| { | |
| "epoch": 2.31629392971246, | |
| "grad_norm": 0.36733072628049745, | |
| "learning_rate": 3.386626950766713e-05, | |
| "loss": 0.1694, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17258095741271973, | |
| "step": 1450, | |
| "valid_targets_mean": 6268.2, | |
| "valid_targets_min": 3690 | |
| }, | |
| { | |
| "epoch": 2.3242811501597442, | |
| "grad_norm": 0.4677614341570051, | |
| "learning_rate": 3.380874280990454e-05, | |
| "loss": 0.1696, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17201688885688782, | |
| "step": 1455, | |
| "valid_targets_mean": 4644.8, | |
| "valid_targets_min": 2486 | |
| }, | |
| { | |
| "epoch": 2.3322683706070286, | |
| "grad_norm": 0.4526492848943914, | |
| "learning_rate": 3.37509969630041e-05, | |
| "loss": 0.1722, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1867167204618454, | |
| "step": 1460, | |
| "valid_targets_mean": 4428.2, | |
| "valid_targets_min": 1335 | |
| }, | |
| { | |
| "epoch": 2.340255591054313, | |
| "grad_norm": 0.403465236236776, | |
| "learning_rate": 3.369303288341078e-05, | |
| "loss": 0.171, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16157183051109314, | |
| "step": 1465, | |
| "valid_targets_mean": 5202.6, | |
| "valid_targets_min": 982 | |
| }, | |
| { | |
| "epoch": 2.3482428115015974, | |
| "grad_norm": 0.4419718618548526, | |
| "learning_rate": 3.363485149103292e-05, | |
| "loss": 0.1606, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1757904291152954, | |
| "step": 1470, | |
| "valid_targets_mean": 5063.3, | |
| "valid_targets_min": 3624 | |
| }, | |
| { | |
| "epoch": 2.356230031948882, | |
| "grad_norm": 0.3912451660748444, | |
| "learning_rate": 3.357645370922772e-05, | |
| "loss": 0.1527, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14932778477668762, | |
| "step": 1475, | |
| "valid_targets_mean": 5627.3, | |
| "valid_targets_min": 3143 | |
| }, | |
| { | |
| "epoch": 2.364217252396166, | |
| "grad_norm": 0.43238174523279177, | |
| "learning_rate": 3.351784046478655e-05, | |
| "loss": 0.1685, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17672443389892578, | |
| "step": 1480, | |
| "valid_targets_mean": 5149.8, | |
| "valid_targets_min": 1925 | |
| }, | |
| { | |
| "epoch": 2.3722044728434506, | |
| "grad_norm": 0.4296657449110657, | |
| "learning_rate": 3.345901268792022e-05, | |
| "loss": 0.1766, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2030010223388672, | |
| "step": 1485, | |
| "valid_targets_mean": 5148.1, | |
| "valid_targets_min": 2365 | |
| }, | |
| { | |
| "epoch": 2.380191693290735, | |
| "grad_norm": 0.41604888915293997, | |
| "learning_rate": 3.3399971312244255e-05, | |
| "loss": 0.1737, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17099419236183167, | |
| "step": 1490, | |
| "valid_targets_mean": 5186.4, | |
| "valid_targets_min": 1953 | |
| }, | |
| { | |
| "epoch": 2.3881789137380194, | |
| "grad_norm": 0.4520192717346139, | |
| "learning_rate": 3.334071727476406e-05, | |
| "loss": 0.1739, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1894477903842926, | |
| "step": 1495, | |
| "valid_targets_mean": 5717.0, | |
| "valid_targets_min": 3014 | |
| }, | |
| { | |
| "epoch": 2.3961661341853033, | |
| "grad_norm": 0.43510439755662034, | |
| "learning_rate": 3.3281251515860035e-05, | |
| "loss": 0.1685, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17742344737052917, | |
| "step": 1500, | |
| "valid_targets_mean": 5522.3, | |
| "valid_targets_min": 2501 | |
| }, | |
| { | |
| "epoch": 2.4041533546325877, | |
| "grad_norm": 0.5594893889136163, | |
| "learning_rate": 3.3221574979272676e-05, | |
| "loss": 0.1673, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18369264900684357, | |
| "step": 1505, | |
| "valid_targets_mean": 5510.2, | |
| "valid_targets_min": 2461 | |
| }, | |
| { | |
| "epoch": 2.412140575079872, | |
| "grad_norm": 0.43111857571278356, | |
| "learning_rate": 3.316168861208759e-05, | |
| "loss": 0.1695, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18407931923866272, | |
| "step": 1510, | |
| "valid_targets_mean": 5056.1, | |
| "valid_targets_min": 3090 | |
| }, | |
| { | |
| "epoch": 2.4201277955271565, | |
| "grad_norm": 0.4118123801700256, | |
| "learning_rate": 3.310159336472047e-05, | |
| "loss": 0.1775, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18072107434272766, | |
| "step": 1515, | |
| "valid_targets_mean": 4679.9, | |
| "valid_targets_min": 1604 | |
| }, | |
| { | |
| "epoch": 2.428115015974441, | |
| "grad_norm": 0.42758731109535736, | |
| "learning_rate": 3.304129019090198e-05, | |
| "loss": 0.1631, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14738571643829346, | |
| "step": 1520, | |
| "valid_targets_mean": 4509.3, | |
| "valid_targets_min": 1565 | |
| }, | |
| { | |
| "epoch": 2.4361022364217253, | |
| "grad_norm": 0.39524566103494335, | |
| "learning_rate": 3.298078004766267e-05, | |
| "loss": 0.1716, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17665189504623413, | |
| "step": 1525, | |
| "valid_targets_mean": 5697.6, | |
| "valid_targets_min": 3143 | |
| }, | |
| { | |
| "epoch": 2.4440894568690097, | |
| "grad_norm": 0.3920883444444094, | |
| "learning_rate": 3.292006389531774e-05, | |
| "loss": 0.1675, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17272454500198364, | |
| "step": 1530, | |
| "valid_targets_mean": 5670.8, | |
| "valid_targets_min": 3701 | |
| }, | |
| { | |
| "epoch": 2.452076677316294, | |
| "grad_norm": 0.4111678679178058, | |
| "learning_rate": 3.2859142697451835e-05, | |
| "loss": 0.1749, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17637260258197784, | |
| "step": 1535, | |
| "valid_targets_mean": 5375.0, | |
| "valid_targets_min": 2262 | |
| }, | |
| { | |
| "epoch": 2.460063897763578, | |
| "grad_norm": 0.4149957655773262, | |
| "learning_rate": 3.279801742090372e-05, | |
| "loss": 0.1558, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13643914461135864, | |
| "step": 1540, | |
| "valid_targets_mean": 4706.2, | |
| "valid_targets_min": 1916 | |
| }, | |
| { | |
| "epoch": 2.4680511182108624, | |
| "grad_norm": 0.39075290244508853, | |
| "learning_rate": 3.2736689035750975e-05, | |
| "loss": 0.1679, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15598708391189575, | |
| "step": 1545, | |
| "valid_targets_mean": 5225.3, | |
| "valid_targets_min": 2760 | |
| }, | |
| { | |
| "epoch": 2.476038338658147, | |
| "grad_norm": 0.4095419115308293, | |
| "learning_rate": 3.2675158515294554e-05, | |
| "loss": 0.1499, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16418322920799255, | |
| "step": 1550, | |
| "valid_targets_mean": 5037.9, | |
| "valid_targets_min": 1515 | |
| }, | |
| { | |
| "epoch": 2.484025559105431, | |
| "grad_norm": 0.41752710181486885, | |
| "learning_rate": 3.2613426836043386e-05, | |
| "loss": 0.1864, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17850902676582336, | |
| "step": 1555, | |
| "valid_targets_mean": 5237.9, | |
| "valid_targets_min": 1554 | |
| }, | |
| { | |
| "epoch": 2.4920127795527156, | |
| "grad_norm": 0.4418170581588643, | |
| "learning_rate": 3.2551494977698844e-05, | |
| "loss": 0.1722, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18662750720977783, | |
| "step": 1560, | |
| "valid_targets_mean": 4776.6, | |
| "valid_targets_min": 2664 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "grad_norm": 0.40913661683811037, | |
| "learning_rate": 3.248936392313921e-05, | |
| "loss": 0.1733, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16212037205696106, | |
| "step": 1565, | |
| "valid_targets_mean": 5119.8, | |
| "valid_targets_min": 1840 | |
| }, | |
| { | |
| "epoch": 2.5079872204472844, | |
| "grad_norm": 0.4260181274038959, | |
| "learning_rate": 3.2427034658404056e-05, | |
| "loss": 0.1534, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15736332535743713, | |
| "step": 1570, | |
| "valid_targets_mean": 4423.1, | |
| "valid_targets_min": 1660 | |
| }, | |
| { | |
| "epoch": 2.515974440894569, | |
| "grad_norm": 0.39522571064905476, | |
| "learning_rate": 3.236450817267863e-05, | |
| "loss": 0.1783, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18762344121932983, | |
| "step": 1575, | |
| "valid_targets_mean": 6161.5, | |
| "valid_targets_min": 2991 | |
| }, | |
| { | |
| "epoch": 2.523961661341853, | |
| "grad_norm": 0.4003544452058309, | |
| "learning_rate": 3.230178545827814e-05, | |
| "loss": 0.1597, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1486572027206421, | |
| "step": 1580, | |
| "valid_targets_mean": 4721.9, | |
| "valid_targets_min": 1657 | |
| }, | |
| { | |
| "epoch": 2.5319488817891376, | |
| "grad_norm": 0.40556016185115784, | |
| "learning_rate": 3.223886751063201e-05, | |
| "loss": 0.1694, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16767382621765137, | |
| "step": 1585, | |
| "valid_targets_mean": 5095.9, | |
| "valid_targets_min": 1868 | |
| }, | |
| { | |
| "epoch": 2.539936102236422, | |
| "grad_norm": 0.40940656088513105, | |
| "learning_rate": 3.2175755328268054e-05, | |
| "loss": 0.1679, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16696970164775848, | |
| "step": 1590, | |
| "valid_targets_mean": 4850.4, | |
| "valid_targets_min": 2696 | |
| }, | |
| { | |
| "epoch": 2.547923322683706, | |
| "grad_norm": 0.39435564000412776, | |
| "learning_rate": 3.2112449912796675e-05, | |
| "loss": 0.1691, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18351686000823975, | |
| "step": 1595, | |
| "valid_targets_mean": 5690.0, | |
| "valid_targets_min": 2806 | |
| }, | |
| { | |
| "epoch": 2.5559105431309903, | |
| "grad_norm": 0.3851418707587981, | |
| "learning_rate": 3.204895226889494e-05, | |
| "loss": 0.1724, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15431320667266846, | |
| "step": 1600, | |
| "valid_targets_mean": 5329.9, | |
| "valid_targets_min": 1827 | |
| }, | |
| { | |
| "epoch": 2.5638977635782747, | |
| "grad_norm": 0.40609003313824465, | |
| "learning_rate": 3.1985263404290634e-05, | |
| "loss": 0.1645, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1583857536315918, | |
| "step": 1605, | |
| "valid_targets_mean": 5451.8, | |
| "valid_targets_min": 1619 | |
| }, | |
| { | |
| "epoch": 2.571884984025559, | |
| "grad_norm": 0.38573429052146013, | |
| "learning_rate": 3.19213843297463e-05, | |
| "loss": 0.183, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1622629165649414, | |
| "step": 1610, | |
| "valid_targets_mean": 5308.6, | |
| "valid_targets_min": 2006 | |
| }, | |
| { | |
| "epoch": 2.5798722044728435, | |
| "grad_norm": 0.4130064715941179, | |
| "learning_rate": 3.1857316059043144e-05, | |
| "loss": 0.1793, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1783134937286377, | |
| "step": 1615, | |
| "valid_targets_mean": 5072.8, | |
| "valid_targets_min": 2276 | |
| }, | |
| { | |
| "epoch": 2.587859424920128, | |
| "grad_norm": 0.4487690042696091, | |
| "learning_rate": 3.1793059608964986e-05, | |
| "loss": 0.1718, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19156934320926666, | |
| "step": 1620, | |
| "valid_targets_mean": 4515.2, | |
| "valid_targets_min": 1496 | |
| }, | |
| { | |
| "epoch": 2.5958466453674123, | |
| "grad_norm": 0.4311649638102354, | |
| "learning_rate": 3.172861599928212e-05, | |
| "loss": 0.1675, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15223652124404907, | |
| "step": 1625, | |
| "valid_targets_mean": 4504.2, | |
| "valid_targets_min": 1432 | |
| }, | |
| { | |
| "epoch": 2.6038338658146962, | |
| "grad_norm": 0.3967433931280785, | |
| "learning_rate": 3.166398625273512e-05, | |
| "loss": 0.1762, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18588952720165253, | |
| "step": 1630, | |
| "valid_targets_mean": 5450.6, | |
| "valid_targets_min": 1232 | |
| }, | |
| { | |
| "epoch": 2.6118210862619806, | |
| "grad_norm": 0.4029445675634546, | |
| "learning_rate": 3.15991713950186e-05, | |
| "loss": 0.1667, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17515186965465546, | |
| "step": 1635, | |
| "valid_targets_mean": 5113.4, | |
| "valid_targets_min": 2651 | |
| }, | |
| { | |
| "epoch": 2.619808306709265, | |
| "grad_norm": 0.40846814040205104, | |
| "learning_rate": 3.153417245476495e-05, | |
| "loss": 0.1615, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17537692189216614, | |
| "step": 1640, | |
| "valid_targets_mean": 4761.1, | |
| "valid_targets_min": 2221 | |
| }, | |
| { | |
| "epoch": 2.6277955271565494, | |
| "grad_norm": 0.46829117503818635, | |
| "learning_rate": 3.146899046352804e-05, | |
| "loss": 0.1661, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15613813698291779, | |
| "step": 1645, | |
| "valid_targets_mean": 5544.8, | |
| "valid_targets_min": 3049 | |
| }, | |
| { | |
| "epoch": 2.635782747603834, | |
| "grad_norm": 0.4168257460858564, | |
| "learning_rate": 3.140362645576676e-05, | |
| "loss": 0.1549, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15841329097747803, | |
| "step": 1650, | |
| "valid_targets_mean": 5226.9, | |
| "valid_targets_min": 2296 | |
| }, | |
| { | |
| "epoch": 2.643769968051118, | |
| "grad_norm": 0.4281719172527959, | |
| "learning_rate": 3.133808146882871e-05, | |
| "loss": 0.171, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18718989193439484, | |
| "step": 1655, | |
| "valid_targets_mean": 5129.5, | |
| "valid_targets_min": 2827 | |
| }, | |
| { | |
| "epoch": 2.6517571884984026, | |
| "grad_norm": 0.4197236148723484, | |
| "learning_rate": 3.1272356542933654e-05, | |
| "loss": 0.1668, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1557224988937378, | |
| "step": 1660, | |
| "valid_targets_mean": 4359.2, | |
| "valid_targets_min": 2883 | |
| }, | |
| { | |
| "epoch": 2.659744408945687, | |
| "grad_norm": 0.4668477596445616, | |
| "learning_rate": 3.120645272115707e-05, | |
| "loss": 0.1734, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18326757848262787, | |
| "step": 1665, | |
| "valid_targets_mean": 4976.5, | |
| "valid_targets_min": 744 | |
| }, | |
| { | |
| "epoch": 2.6677316293929714, | |
| "grad_norm": 0.4480600891396075, | |
| "learning_rate": 3.114037104941355e-05, | |
| "loss": 0.1772, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20831947028636932, | |
| "step": 1670, | |
| "valid_targets_mean": 5018.0, | |
| "valid_targets_min": 967 | |
| }, | |
| { | |
| "epoch": 2.6757188498402558, | |
| "grad_norm": 0.38296249450478514, | |
| "learning_rate": 3.1074112576440236e-05, | |
| "loss": 0.1724, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15984682738780975, | |
| "step": 1675, | |
| "valid_targets_mean": 5199.9, | |
| "valid_targets_min": 2335 | |
| }, | |
| { | |
| "epoch": 2.68370607028754, | |
| "grad_norm": 0.3834588488635434, | |
| "learning_rate": 3.1007678353780154e-05, | |
| "loss": 0.1647, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16989043354988098, | |
| "step": 1680, | |
| "valid_targets_mean": 5081.8, | |
| "valid_targets_min": 1725 | |
| }, | |
| { | |
| "epoch": 2.6916932907348246, | |
| "grad_norm": 0.41954023319021827, | |
| "learning_rate": 3.094106943576553e-05, | |
| "loss": 0.1522, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16213415563106537, | |
| "step": 1685, | |
| "valid_targets_mean": 4986.4, | |
| "valid_targets_min": 1987 | |
| }, | |
| { | |
| "epoch": 2.6996805111821085, | |
| "grad_norm": 0.4335320537803012, | |
| "learning_rate": 3.087428687950108e-05, | |
| "loss": 0.1643, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1695536971092224, | |
| "step": 1690, | |
| "valid_targets_mean": 4851.8, | |
| "valid_targets_min": 2235 | |
| }, | |
| { | |
| "epoch": 2.707667731629393, | |
| "grad_norm": 0.3598621208938739, | |
| "learning_rate": 3.080733174484719e-05, | |
| "loss": 0.1792, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16863268613815308, | |
| "step": 1695, | |
| "valid_targets_mean": 5683.1, | |
| "valid_targets_min": 2548 | |
| }, | |
| { | |
| "epoch": 2.7156549520766773, | |
| "grad_norm": 0.4119830513324859, | |
| "learning_rate": 3.074020509440313e-05, | |
| "loss": 0.1565, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16991566121578217, | |
| "step": 1700, | |
| "valid_targets_mean": 5297.9, | |
| "valid_targets_min": 1838 | |
| }, | |
| { | |
| "epoch": 2.7236421725239617, | |
| "grad_norm": 0.41087132009895544, | |
| "learning_rate": 3.0672907993490183e-05, | |
| "loss": 0.1747, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24218526482582092, | |
| "step": 1705, | |
| "valid_targets_mean": 5788.6, | |
| "valid_targets_min": 2880 | |
| }, | |
| { | |
| "epoch": 2.731629392971246, | |
| "grad_norm": 0.4206009407868746, | |
| "learning_rate": 3.060544151013473e-05, | |
| "loss": 0.162, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1875559687614441, | |
| "step": 1710, | |
| "valid_targets_mean": 5178.4, | |
| "valid_targets_min": 2166 | |
| }, | |
| { | |
| "epoch": 2.7396166134185305, | |
| "grad_norm": 0.4557066957251873, | |
| "learning_rate": 3.053780671505132e-05, | |
| "loss": 0.155, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1581772416830063, | |
| "step": 1715, | |
| "valid_targets_mean": 4518.5, | |
| "valid_targets_min": 2139 | |
| }, | |
| { | |
| "epoch": 2.747603833865815, | |
| "grad_norm": 0.3925588195510952, | |
| "learning_rate": 3.047000468162563e-05, | |
| "loss": 0.1681, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1605645716190338, | |
| "step": 1720, | |
| "valid_targets_mean": 4778.3, | |
| "valid_targets_min": 1954 | |
| }, | |
| { | |
| "epoch": 2.755591054313099, | |
| "grad_norm": 0.40508343542097747, | |
| "learning_rate": 3.0402036485897496e-05, | |
| "loss": 0.1719, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1733085811138153, | |
| "step": 1725, | |
| "valid_targets_mean": 4970.4, | |
| "valid_targets_min": 622 | |
| }, | |
| { | |
| "epoch": 2.763578274760383, | |
| "grad_norm": 0.45310224562229384, | |
| "learning_rate": 3.0333903206543786e-05, | |
| "loss": 0.1727, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19950124621391296, | |
| "step": 1730, | |
| "valid_targets_mean": 4365.4, | |
| "valid_targets_min": 597 | |
| }, | |
| { | |
| "epoch": 2.7715654952076676, | |
| "grad_norm": 0.48189506452475217, | |
| "learning_rate": 3.02656059248613e-05, | |
| "loss": 0.1599, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16813474893569946, | |
| "step": 1735, | |
| "valid_targets_mean": 5012.3, | |
| "valid_targets_min": 2442 | |
| }, | |
| { | |
| "epoch": 2.779552715654952, | |
| "grad_norm": 0.3672138597807966, | |
| "learning_rate": 3.0197145724749615e-05, | |
| "loss": 0.1742, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14925971627235413, | |
| "step": 1740, | |
| "valid_targets_mean": 5605.8, | |
| "valid_targets_min": 2871 | |
| }, | |
| { | |
| "epoch": 2.7875399361022364, | |
| "grad_norm": 0.4470534331488075, | |
| "learning_rate": 3.0128523692693865e-05, | |
| "loss": 0.1718, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1754777878522873, | |
| "step": 1745, | |
| "valid_targets_mean": 4590.7, | |
| "valid_targets_min": 1129 | |
| }, | |
| { | |
| "epoch": 2.7955271565495208, | |
| "grad_norm": 0.3856037514237977, | |
| "learning_rate": 3.005974091774751e-05, | |
| "loss": 0.1724, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1683935970067978, | |
| "step": 1750, | |
| "valid_targets_mean": 4804.4, | |
| "valid_targets_min": 2376 | |
| }, | |
| { | |
| "epoch": 2.803514376996805, | |
| "grad_norm": 0.3835900248880264, | |
| "learning_rate": 2.999079849151505e-05, | |
| "loss": 0.1776, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17427316308021545, | |
| "step": 1755, | |
| "valid_targets_mean": 5613.7, | |
| "valid_targets_min": 2850 | |
| }, | |
| { | |
| "epoch": 2.8115015974440896, | |
| "grad_norm": 0.4489164835062511, | |
| "learning_rate": 2.992169750813471e-05, | |
| "loss": 0.1697, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22016024589538574, | |
| "step": 1760, | |
| "valid_targets_mean": 5062.6, | |
| "valid_targets_min": 2442 | |
| }, | |
| { | |
| "epoch": 2.819488817891374, | |
| "grad_norm": 0.4438065060291003, | |
| "learning_rate": 2.985243906426106e-05, | |
| "loss": 0.1725, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1704789400100708, | |
| "step": 1765, | |
| "valid_targets_mean": 4789.8, | |
| "valid_targets_min": 2814 | |
| }, | |
| { | |
| "epoch": 2.8274760383386583, | |
| "grad_norm": 0.4536197101306054, | |
| "learning_rate": 2.9783024259047617e-05, | |
| "loss": 0.1746, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17780658602714539, | |
| "step": 1770, | |
| "valid_targets_mean": 4168.8, | |
| "valid_targets_min": 2026 | |
| }, | |
| { | |
| "epoch": 2.8354632587859427, | |
| "grad_norm": 0.4299937345910055, | |
| "learning_rate": 2.971345419412941e-05, | |
| "loss": 0.174, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17868943512439728, | |
| "step": 1775, | |
| "valid_targets_mean": 4880.2, | |
| "valid_targets_min": 1180 | |
| }, | |
| { | |
| "epoch": 2.8434504792332267, | |
| "grad_norm": 0.4193736023016212, | |
| "learning_rate": 2.964372997360548e-05, | |
| "loss": 0.1653, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17995405197143555, | |
| "step": 1780, | |
| "valid_targets_mean": 5075.3, | |
| "valid_targets_min": 1975 | |
| }, | |
| { | |
| "epoch": 2.851437699680511, | |
| "grad_norm": 0.3972100967099474, | |
| "learning_rate": 2.957385270402137e-05, | |
| "loss": 0.1674, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16427266597747803, | |
| "step": 1785, | |
| "valid_targets_mean": 4612.9, | |
| "valid_targets_min": 1065 | |
| }, | |
| { | |
| "epoch": 2.8594249201277955, | |
| "grad_norm": 0.39866837086375273, | |
| "learning_rate": 2.9503823494351565e-05, | |
| "loss": 0.1611, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16262301802635193, | |
| "step": 1790, | |
| "valid_targets_mean": 4942.6, | |
| "valid_targets_min": 2014 | |
| }, | |
| { | |
| "epoch": 2.86741214057508, | |
| "grad_norm": 0.40434476358683685, | |
| "learning_rate": 2.9433643455981874e-05, | |
| "loss": 0.1773, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17950189113616943, | |
| "step": 1795, | |
| "valid_targets_mean": 5180.8, | |
| "valid_targets_min": 3425 | |
| }, | |
| { | |
| "epoch": 2.8753993610223643, | |
| "grad_norm": 0.4253565523791667, | |
| "learning_rate": 2.9363313702691827e-05, | |
| "loss": 0.1702, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19628484547138214, | |
| "step": 1800, | |
| "valid_targets_mean": 4755.3, | |
| "valid_targets_min": 1016 | |
| }, | |
| { | |
| "epoch": 2.8833865814696487, | |
| "grad_norm": 0.4123065342308681, | |
| "learning_rate": 2.9292835350636957e-05, | |
| "loss": 0.1621, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1509486436843872, | |
| "step": 1805, | |
| "valid_targets_mean": 5108.9, | |
| "valid_targets_min": 1683 | |
| }, | |
| { | |
| "epoch": 2.891373801916933, | |
| "grad_norm": 0.42595879606896164, | |
| "learning_rate": 2.922220951833111e-05, | |
| "loss": 0.1653, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1909659057855606, | |
| "step": 1810, | |
| "valid_targets_mean": 4969.5, | |
| "valid_targets_min": 2661 | |
| }, | |
| { | |
| "epoch": 2.899361022364217, | |
| "grad_norm": 0.45101856509575244, | |
| "learning_rate": 2.9151437326628706e-05, | |
| "loss": 0.1609, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15321360528469086, | |
| "step": 1815, | |
| "valid_targets_mean": 4815.4, | |
| "valid_targets_min": 2862 | |
| }, | |
| { | |
| "epoch": 2.9073482428115014, | |
| "grad_norm": 0.40173840543814443, | |
| "learning_rate": 2.908051989870692e-05, | |
| "loss": 0.1634, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19088567793369293, | |
| "step": 1820, | |
| "valid_targets_mean": 5024.5, | |
| "valid_targets_min": 1588 | |
| }, | |
| { | |
| "epoch": 2.915335463258786, | |
| "grad_norm": 0.40329358599652815, | |
| "learning_rate": 2.9009458360047867e-05, | |
| "loss": 0.1718, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20809268951416016, | |
| "step": 1825, | |
| "valid_targets_mean": 5024.4, | |
| "valid_targets_min": 2343 | |
| }, | |
| { | |
| "epoch": 2.92332268370607, | |
| "grad_norm": 0.41813866789626447, | |
| "learning_rate": 2.8938253838420762e-05, | |
| "loss": 0.1651, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1576555222272873, | |
| "step": 1830, | |
| "valid_targets_mean": 5563.0, | |
| "valid_targets_min": 2735 | |
| }, | |
| { | |
| "epoch": 2.9313099041533546, | |
| "grad_norm": 0.4284051121724335, | |
| "learning_rate": 2.8866907463864006e-05, | |
| "loss": 0.1598, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1685807853937149, | |
| "step": 1835, | |
| "valid_targets_mean": 4706.4, | |
| "valid_targets_min": 866 | |
| }, | |
| { | |
| "epoch": 2.939297124600639, | |
| "grad_norm": 0.4117023523138034, | |
| "learning_rate": 2.8795420368667225e-05, | |
| "loss": 0.1627, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16330750286579132, | |
| "step": 1840, | |
| "valid_targets_mean": 4569.9, | |
| "valid_targets_min": 2268 | |
| }, | |
| { | |
| "epoch": 2.9472843450479234, | |
| "grad_norm": 0.4692337582745295, | |
| "learning_rate": 2.8723793687353354e-05, | |
| "loss": 0.1727, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16872987151145935, | |
| "step": 1845, | |
| "valid_targets_mean": 4526.9, | |
| "valid_targets_min": 2444 | |
| }, | |
| { | |
| "epoch": 2.9552715654952078, | |
| "grad_norm": 0.41736414193536114, | |
| "learning_rate": 2.8652028556660593e-05, | |
| "loss": 0.17, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19827672839164734, | |
| "step": 1850, | |
| "valid_targets_mean": 5740.1, | |
| "valid_targets_min": 2462 | |
| }, | |
| { | |
| "epoch": 2.963258785942492, | |
| "grad_norm": 0.4272587476160979, | |
| "learning_rate": 2.8580126115524373e-05, | |
| "loss": 0.1659, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16046331822872162, | |
| "step": 1855, | |
| "valid_targets_mean": 4532.7, | |
| "valid_targets_min": 601 | |
| }, | |
| { | |
| "epoch": 2.9712460063897765, | |
| "grad_norm": 0.4119433783254139, | |
| "learning_rate": 2.8508087505059302e-05, | |
| "loss": 0.1599, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16485300660133362, | |
| "step": 1860, | |
| "valid_targets_mean": 5264.5, | |
| "valid_targets_min": 2163 | |
| }, | |
| { | |
| "epoch": 2.979233226837061, | |
| "grad_norm": 0.41296556451702676, | |
| "learning_rate": 2.843591386854102e-05, | |
| "loss": 0.1681, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16444364190101624, | |
| "step": 1865, | |
| "valid_targets_mean": 4693.9, | |
| "valid_targets_min": 1124 | |
| }, | |
| { | |
| "epoch": 2.987220447284345, | |
| "grad_norm": 0.4230058344287007, | |
| "learning_rate": 2.8363606351388068e-05, | |
| "loss": 0.1614, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18372346460819244, | |
| "step": 1870, | |
| "valid_targets_mean": 4846.9, | |
| "valid_targets_min": 1222 | |
| }, | |
| { | |
| "epoch": 2.9952076677316293, | |
| "grad_norm": 0.41695362931000385, | |
| "learning_rate": 2.829116610114375e-05, | |
| "loss": 0.1673, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18323427438735962, | |
| "step": 1875, | |
| "valid_targets_mean": 5672.2, | |
| "valid_targets_min": 2033 | |
| }, | |
| { | |
| "epoch": 3.0031948881789137, | |
| "grad_norm": 0.39416435317649495, | |
| "learning_rate": 2.821859426745786e-05, | |
| "loss": 0.1674, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14279429614543915, | |
| "step": 1880, | |
| "valid_targets_mean": 4793.4, | |
| "valid_targets_min": 2968 | |
| }, | |
| { | |
| "epoch": 3.011182108626198, | |
| "grad_norm": 0.4912779372362586, | |
| "learning_rate": 2.8145892002068454e-05, | |
| "loss": 0.1523, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14633120596408844, | |
| "step": 1885, | |
| "valid_targets_mean": 4848.0, | |
| "valid_targets_min": 1390 | |
| }, | |
| { | |
| "epoch": 3.0191693290734825, | |
| "grad_norm": 0.3615730973602941, | |
| "learning_rate": 2.8073060458783606e-05, | |
| "loss": 0.1538, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14516273140907288, | |
| "step": 1890, | |
| "valid_targets_mean": 5678.5, | |
| "valid_targets_min": 434 | |
| }, | |
| { | |
| "epoch": 3.027156549520767, | |
| "grad_norm": 0.45593893636684485, | |
| "learning_rate": 2.8000100793463056e-05, | |
| "loss": 0.1614, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15923550724983215, | |
| "step": 1895, | |
| "valid_targets_mean": 5171.6, | |
| "valid_targets_min": 1946 | |
| }, | |
| { | |
| "epoch": 3.0351437699680512, | |
| "grad_norm": 0.3911719531842014, | |
| "learning_rate": 2.792701416399988e-05, | |
| "loss": 0.1636, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14421124756336212, | |
| "step": 1900, | |
| "valid_targets_mean": 5711.9, | |
| "valid_targets_min": 2778 | |
| }, | |
| { | |
| "epoch": 3.0431309904153356, | |
| "grad_norm": 0.38307179381210565, | |
| "learning_rate": 2.7853801730302134e-05, | |
| "loss": 0.1525, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16137105226516724, | |
| "step": 1905, | |
| "valid_targets_mean": 5577.6, | |
| "valid_targets_min": 2061 | |
| }, | |
| { | |
| "epoch": 3.0511182108626196, | |
| "grad_norm": 0.41008114821659913, | |
| "learning_rate": 2.7780464654274396e-05, | |
| "loss": 0.1546, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13742196559906006, | |
| "step": 1910, | |
| "valid_targets_mean": 5102.4, | |
| "valid_targets_min": 2979 | |
| }, | |
| { | |
| "epoch": 3.059105431309904, | |
| "grad_norm": 0.4307363059969895, | |
| "learning_rate": 2.770700409979938e-05, | |
| "loss": 0.14, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15213440358638763, | |
| "step": 1915, | |
| "valid_targets_mean": 5177.0, | |
| "valid_targets_min": 2162 | |
| }, | |
| { | |
| "epoch": 3.0670926517571884, | |
| "grad_norm": 0.42338944432300873, | |
| "learning_rate": 2.7633421232719442e-05, | |
| "loss": 0.1594, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1825188398361206, | |
| "step": 1920, | |
| "valid_targets_mean": 5494.0, | |
| "valid_targets_min": 1112 | |
| }, | |
| { | |
| "epoch": 3.0750798722044728, | |
| "grad_norm": 0.40678108024424003, | |
| "learning_rate": 2.7559717220818067e-05, | |
| "loss": 0.1517, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17005711793899536, | |
| "step": 1925, | |
| "valid_targets_mean": 5216.4, | |
| "valid_targets_min": 2444 | |
| }, | |
| { | |
| "epoch": 3.083067092651757, | |
| "grad_norm": 0.4174201639190384, | |
| "learning_rate": 2.7485893233801366e-05, | |
| "loss": 0.1437, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14175830781459808, | |
| "step": 1930, | |
| "valid_targets_mean": 4799.2, | |
| "valid_targets_min": 894 | |
| }, | |
| { | |
| "epoch": 3.0910543130990416, | |
| "grad_norm": 0.4212032001789445, | |
| "learning_rate": 2.7411950443279458e-05, | |
| "loss": 0.1502, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14395883679389954, | |
| "step": 1935, | |
| "valid_targets_mean": 5277.1, | |
| "valid_targets_min": 3718 | |
| }, | |
| { | |
| "epoch": 3.099041533546326, | |
| "grad_norm": 0.5734241719912586, | |
| "learning_rate": 2.7337890022747945e-05, | |
| "loss": 0.156, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1667754054069519, | |
| "step": 1940, | |
| "valid_targets_mean": 4890.9, | |
| "valid_targets_min": 748 | |
| }, | |
| { | |
| "epoch": 3.1070287539936103, | |
| "grad_norm": 0.4614951252278709, | |
| "learning_rate": 2.7263713147569243e-05, | |
| "loss": 0.161, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1640327274799347, | |
| "step": 1945, | |
| "valid_targets_mean": 5194.8, | |
| "valid_targets_min": 3292 | |
| }, | |
| { | |
| "epoch": 3.1150159744408947, | |
| "grad_norm": 0.3998166118113714, | |
| "learning_rate": 2.7189420994953945e-05, | |
| "loss": 0.1426, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1297355741262436, | |
| "step": 1950, | |
| "valid_targets_mean": 4903.6, | |
| "valid_targets_min": 3317 | |
| }, | |
| { | |
| "epoch": 3.123003194888179, | |
| "grad_norm": 0.4031662591421517, | |
| "learning_rate": 2.7115014743942114e-05, | |
| "loss": 0.1411, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1414220780134201, | |
| "step": 1955, | |
| "valid_targets_mean": 4659.1, | |
| "valid_targets_min": 1048 | |
| }, | |
| { | |
| "epoch": 3.130990415335463, | |
| "grad_norm": 0.46449312759467615, | |
| "learning_rate": 2.704049557538461e-05, | |
| "loss": 0.1588, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17587429285049438, | |
| "step": 1960, | |
| "valid_targets_mean": 4656.2, | |
| "valid_targets_min": 1660 | |
| }, | |
| { | |
| "epoch": 3.1389776357827475, | |
| "grad_norm": 0.44130220393154124, | |
| "learning_rate": 2.696586467192434e-05, | |
| "loss": 0.1631, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14803341031074524, | |
| "step": 1965, | |
| "valid_targets_mean": 4765.4, | |
| "valid_targets_min": 2164 | |
| }, | |
| { | |
| "epoch": 3.146964856230032, | |
| "grad_norm": 0.4285102614456115, | |
| "learning_rate": 2.6891123217977448e-05, | |
| "loss": 0.154, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12965035438537598, | |
| "step": 1970, | |
| "valid_targets_mean": 4780.1, | |
| "valid_targets_min": 2543 | |
| }, | |
| { | |
| "epoch": 3.1549520766773163, | |
| "grad_norm": 0.4044200880892866, | |
| "learning_rate": 2.681627239971458e-05, | |
| "loss": 0.1585, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15685327351093292, | |
| "step": 1975, | |
| "valid_targets_mean": 5372.6, | |
| "valid_targets_min": 2394 | |
| }, | |
| { | |
| "epoch": 3.1629392971246006, | |
| "grad_norm": 0.397635209029452, | |
| "learning_rate": 2.6741313405041997e-05, | |
| "loss": 0.1484, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14557038247585297, | |
| "step": 1980, | |
| "valid_targets_mean": 4826.5, | |
| "valid_targets_min": 2040 | |
| }, | |
| { | |
| "epoch": 3.170926517571885, | |
| "grad_norm": 0.44982277889108496, | |
| "learning_rate": 2.666624742358279e-05, | |
| "loss": 0.1578, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15522637963294983, | |
| "step": 1985, | |
| "valid_targets_mean": 4646.1, | |
| "valid_targets_min": 1659 | |
| }, | |
| { | |
| "epoch": 3.1789137380191694, | |
| "grad_norm": 0.4369201664276143, | |
| "learning_rate": 2.659107564665794e-05, | |
| "loss": 0.1504, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15977069735527039, | |
| "step": 1990, | |
| "valid_targets_mean": 5445.1, | |
| "valid_targets_min": 1570 | |
| }, | |
| { | |
| "epoch": 3.186900958466454, | |
| "grad_norm": 0.5012526558376844, | |
| "learning_rate": 2.6515799267267445e-05, | |
| "loss": 0.1506, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14220066368579865, | |
| "step": 1995, | |
| "valid_targets_mean": 4817.1, | |
| "valid_targets_min": 2060 | |
| }, | |
| { | |
| "epoch": 3.194888178913738, | |
| "grad_norm": 0.4144195970719363, | |
| "learning_rate": 2.6440419480071366e-05, | |
| "loss": 0.1564, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.145687535405159, | |
| "step": 2000, | |
| "valid_targets_mean": 4871.1, | |
| "valid_targets_min": 1685 | |
| }, | |
| { | |
| "epoch": 3.202875399361022, | |
| "grad_norm": 0.42788490157736064, | |
| "learning_rate": 2.6364937481370887e-05, | |
| "loss": 0.1685, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17767146229743958, | |
| "step": 2005, | |
| "valid_targets_mean": 5114.2, | |
| "valid_targets_min": 2028 | |
| }, | |
| { | |
| "epoch": 3.2108626198083066, | |
| "grad_norm": 0.4313021433183399, | |
| "learning_rate": 2.628935446908933e-05, | |
| "loss": 0.1455, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14824821054935455, | |
| "step": 2010, | |
| "valid_targets_mean": 4867.8, | |
| "valid_targets_min": 2676 | |
| }, | |
| { | |
| "epoch": 3.218849840255591, | |
| "grad_norm": 0.41576702864903514, | |
| "learning_rate": 2.6213671642753124e-05, | |
| "loss": 0.1543, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1736057996749878, | |
| "step": 2015, | |
| "valid_targets_mean": 5254.3, | |
| "valid_targets_min": 2850 | |
| }, | |
| { | |
| "epoch": 3.2268370607028753, | |
| "grad_norm": 0.4167309067927588, | |
| "learning_rate": 2.6137890203472786e-05, | |
| "loss": 0.1548, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1459808647632599, | |
| "step": 2020, | |
| "valid_targets_mean": 5010.1, | |
| "valid_targets_min": 2475 | |
| }, | |
| { | |
| "epoch": 3.2348242811501597, | |
| "grad_norm": 0.3963753159014156, | |
| "learning_rate": 2.606201135392383e-05, | |
| "loss": 0.1446, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14964553713798523, | |
| "step": 2025, | |
| "valid_targets_mean": 5607.8, | |
| "valid_targets_min": 2779 | |
| }, | |
| { | |
| "epoch": 3.242811501597444, | |
| "grad_norm": 0.475666876826538, | |
| "learning_rate": 2.5986036298327725e-05, | |
| "loss": 0.15, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19719651341438293, | |
| "step": 2030, | |
| "valid_targets_mean": 5536.4, | |
| "valid_targets_min": 336 | |
| }, | |
| { | |
| "epoch": 3.2507987220447285, | |
| "grad_norm": 0.4080066781664306, | |
| "learning_rate": 2.590996624243276e-05, | |
| "loss": 0.1628, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1386992484331131, | |
| "step": 2035, | |
| "valid_targets_mean": 4367.1, | |
| "valid_targets_min": 1718 | |
| }, | |
| { | |
| "epoch": 3.258785942492013, | |
| "grad_norm": 0.4335384113565266, | |
| "learning_rate": 2.5833802393494897e-05, | |
| "loss": 0.1461, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15026232600212097, | |
| "step": 2040, | |
| "valid_targets_mean": 5106.6, | |
| "valid_targets_min": 2391 | |
| }, | |
| { | |
| "epoch": 3.2667731629392973, | |
| "grad_norm": 0.40064444286028195, | |
| "learning_rate": 2.5757545960258638e-05, | |
| "loss": 0.1466, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14546141028404236, | |
| "step": 2045, | |
| "valid_targets_mean": 4911.0, | |
| "valid_targets_min": 2956 | |
| }, | |
| { | |
| "epoch": 3.2747603833865817, | |
| "grad_norm": 0.43939977886681336, | |
| "learning_rate": 2.5681198152937804e-05, | |
| "loss": 0.1536, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16095392405986786, | |
| "step": 2050, | |
| "valid_targets_mean": 5673.7, | |
| "valid_targets_min": 3105 | |
| }, | |
| { | |
| "epoch": 3.2827476038338657, | |
| "grad_norm": 0.39827163101423124, | |
| "learning_rate": 2.5604760183196377e-05, | |
| "loss": 0.1669, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17011982202529907, | |
| "step": 2055, | |
| "valid_targets_mean": 5624.6, | |
| "valid_targets_min": 2399 | |
| }, | |
| { | |
| "epoch": 3.29073482428115, | |
| "grad_norm": 0.3975595194878945, | |
| "learning_rate": 2.5528233264129228e-05, | |
| "loss": 0.1468, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14791333675384521, | |
| "step": 2060, | |
| "valid_targets_mean": 5106.6, | |
| "valid_targets_min": 3085 | |
| }, | |
| { | |
| "epoch": 3.2987220447284344, | |
| "grad_norm": 0.44174367282959814, | |
| "learning_rate": 2.54516186102429e-05, | |
| "loss": 0.1594, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16217058897018433, | |
| "step": 2065, | |
| "valid_targets_mean": 5234.4, | |
| "valid_targets_min": 2650 | |
| }, | |
| { | |
| "epoch": 3.306709265175719, | |
| "grad_norm": 0.44650544101188777, | |
| "learning_rate": 2.5374917437436294e-05, | |
| "loss": 0.1531, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15554942190647125, | |
| "step": 2070, | |
| "valid_targets_mean": 4268.9, | |
| "valid_targets_min": 1685 | |
| }, | |
| { | |
| "epoch": 3.3146964856230032, | |
| "grad_norm": 0.3931600691185933, | |
| "learning_rate": 2.5298130962981403e-05, | |
| "loss": 0.1623, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15588781237602234, | |
| "step": 2075, | |
| "valid_targets_mean": 5491.3, | |
| "valid_targets_min": 3190 | |
| }, | |
| { | |
| "epoch": 3.3226837060702876, | |
| "grad_norm": 0.4198383012921021, | |
| "learning_rate": 2.5221260405503985e-05, | |
| "loss": 0.1605, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1660890281200409, | |
| "step": 2080, | |
| "valid_targets_mean": 4900.6, | |
| "valid_targets_min": 1032 | |
| }, | |
| { | |
| "epoch": 3.330670926517572, | |
| "grad_norm": 0.40480077020809957, | |
| "learning_rate": 2.5144306984964225e-05, | |
| "loss": 0.1437, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16281050443649292, | |
| "step": 2085, | |
| "valid_targets_mean": 5368.4, | |
| "valid_targets_min": 3050 | |
| }, | |
| { | |
| "epoch": 3.3386581469648564, | |
| "grad_norm": 0.44780814837203187, | |
| "learning_rate": 2.5067271922637377e-05, | |
| "loss": 0.1472, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13795217871665955, | |
| "step": 2090, | |
| "valid_targets_mean": 4916.5, | |
| "valid_targets_min": 839 | |
| }, | |
| { | |
| "epoch": 3.3466453674121404, | |
| "grad_norm": 0.39931914642038835, | |
| "learning_rate": 2.499015644109435e-05, | |
| "loss": 0.1378, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13646557927131653, | |
| "step": 2095, | |
| "valid_targets_mean": 5351.0, | |
| "valid_targets_min": 2822 | |
| }, | |
| { | |
| "epoch": 3.3546325878594248, | |
| "grad_norm": 0.5021658943802342, | |
| "learning_rate": 2.491296176418236e-05, | |
| "loss": 0.1524, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1597982943058014, | |
| "step": 2100, | |
| "valid_targets_mean": 4528.8, | |
| "valid_targets_min": 563 | |
| }, | |
| { | |
| "epoch": 3.362619808306709, | |
| "grad_norm": 0.3841944941398438, | |
| "learning_rate": 2.4835689117005472e-05, | |
| "loss": 0.1619, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13390065729618073, | |
| "step": 2105, | |
| "valid_targets_mean": 5489.8, | |
| "valid_targets_min": 868 | |
| }, | |
| { | |
| "epoch": 3.3706070287539935, | |
| "grad_norm": 0.37436494959919714, | |
| "learning_rate": 2.4758339725905152e-05, | |
| "loss": 0.1502, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13600534200668335, | |
| "step": 2110, | |
| "valid_targets_mean": 5530.8, | |
| "valid_targets_min": 1383 | |
| }, | |
| { | |
| "epoch": 3.378594249201278, | |
| "grad_norm": 0.4169837591735533, | |
| "learning_rate": 2.4680914818440823e-05, | |
| "loss": 0.1494, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13024669885635376, | |
| "step": 2115, | |
| "valid_targets_mean": 5691.2, | |
| "valid_targets_min": 3251 | |
| }, | |
| { | |
| "epoch": 3.3865814696485623, | |
| "grad_norm": 0.42977408477835816, | |
| "learning_rate": 2.4603415623370387e-05, | |
| "loss": 0.1561, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16091200709342957, | |
| "step": 2120, | |
| "valid_targets_mean": 5129.2, | |
| "valid_targets_min": 3067 | |
| }, | |
| { | |
| "epoch": 3.3945686900958467, | |
| "grad_norm": 0.4241859454957953, | |
| "learning_rate": 2.4525843370630697e-05, | |
| "loss": 0.1513, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1583596169948578, | |
| "step": 2125, | |
| "valid_targets_mean": 5237.9, | |
| "valid_targets_min": 1714 | |
| }, | |
| { | |
| "epoch": 3.402555910543131, | |
| "grad_norm": 0.6754663173090817, | |
| "learning_rate": 2.4448199291318058e-05, | |
| "loss": 0.1615, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.133111834526062, | |
| "step": 2130, | |
| "valid_targets_mean": 4781.8, | |
| "valid_targets_min": 2253 | |
| }, | |
| { | |
| "epoch": 3.4105431309904155, | |
| "grad_norm": 0.40068171658964125, | |
| "learning_rate": 2.4370484617668707e-05, | |
| "loss": 0.1534, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1510893702507019, | |
| "step": 2135, | |
| "valid_targets_mean": 5122.5, | |
| "valid_targets_min": 2177 | |
| }, | |
| { | |
| "epoch": 3.4185303514377, | |
| "grad_norm": 0.4451562736477082, | |
| "learning_rate": 2.4292700583039194e-05, | |
| "loss": 0.1582, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17255088686943054, | |
| "step": 2140, | |
| "valid_targets_mean": 4662.8, | |
| "valid_targets_min": 2109 | |
| }, | |
| { | |
| "epoch": 3.426517571884984, | |
| "grad_norm": 0.3893655905541626, | |
| "learning_rate": 2.4214848421886893e-05, | |
| "loss": 0.1554, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12814739346504211, | |
| "step": 2145, | |
| "valid_targets_mean": 5397.1, | |
| "valid_targets_min": 2525 | |
| }, | |
| { | |
| "epoch": 3.4345047923322682, | |
| "grad_norm": 0.4180108411154039, | |
| "learning_rate": 2.4136929369750344e-05, | |
| "loss": 0.1599, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13878056406974792, | |
| "step": 2150, | |
| "valid_targets_mean": 4689.8, | |
| "valid_targets_min": 1124 | |
| }, | |
| { | |
| "epoch": 3.4424920127795526, | |
| "grad_norm": 0.37217745786843465, | |
| "learning_rate": 2.4058944663229672e-05, | |
| "loss": 0.1496, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13813605904579163, | |
| "step": 2155, | |
| "valid_targets_mean": 5400.2, | |
| "valid_targets_min": 2245 | |
| }, | |
| { | |
| "epoch": 3.450479233226837, | |
| "grad_norm": 0.39234261091567263, | |
| "learning_rate": 2.398089553996696e-05, | |
| "loss": 0.161, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1626664698123932, | |
| "step": 2160, | |
| "valid_targets_mean": 5755.6, | |
| "valid_targets_min": 2781 | |
| }, | |
| { | |
| "epoch": 3.4584664536741214, | |
| "grad_norm": 0.4267247690566104, | |
| "learning_rate": 2.3902783238626604e-05, | |
| "loss": 0.1446, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14847370982170105, | |
| "step": 2165, | |
| "valid_targets_mean": 4688.8, | |
| "valid_targets_min": 2552 | |
| }, | |
| { | |
| "epoch": 3.466453674121406, | |
| "grad_norm": 0.39538145969151994, | |
| "learning_rate": 2.3824608998875652e-05, | |
| "loss": 0.1459, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1346844732761383, | |
| "step": 2170, | |
| "valid_targets_mean": 4710.8, | |
| "valid_targets_min": 1486 | |
| }, | |
| { | |
| "epoch": 3.47444089456869, | |
| "grad_norm": 0.4314192670103845, | |
| "learning_rate": 2.3746374061364138e-05, | |
| "loss": 0.1707, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14241887629032135, | |
| "step": 2175, | |
| "valid_targets_mean": 4620.2, | |
| "valid_targets_min": 2050 | |
| }, | |
| { | |
| "epoch": 3.4824281150159746, | |
| "grad_norm": 0.5036463928877274, | |
| "learning_rate": 2.3668079667705404e-05, | |
| "loss": 0.1664, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14716178178787231, | |
| "step": 2180, | |
| "valid_targets_mean": 4520.2, | |
| "valid_targets_min": 1051 | |
| }, | |
| { | |
| "epoch": 3.4904153354632586, | |
| "grad_norm": 0.46946026724129697, | |
| "learning_rate": 2.3589727060456342e-05, | |
| "loss": 0.1491, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14512842893600464, | |
| "step": 2185, | |
| "valid_targets_mean": 4685.9, | |
| "valid_targets_min": 1815 | |
| }, | |
| { | |
| "epoch": 3.498402555910543, | |
| "grad_norm": 0.4626432142004945, | |
| "learning_rate": 2.3511317483097753e-05, | |
| "loss": 0.145, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16195783019065857, | |
| "step": 2190, | |
| "valid_targets_mean": 4550.7, | |
| "valid_targets_min": 1780 | |
| }, | |
| { | |
| "epoch": 3.5063897763578273, | |
| "grad_norm": 0.6925254557284592, | |
| "learning_rate": 2.3432852180014548e-05, | |
| "loss": 0.156, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1405593752861023, | |
| "step": 2195, | |
| "valid_targets_mean": 5128.8, | |
| "valid_targets_min": 2404 | |
| }, | |
| { | |
| "epoch": 3.5143769968051117, | |
| "grad_norm": 0.421889156096018, | |
| "learning_rate": 2.3354332396476026e-05, | |
| "loss": 0.1545, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17455607652664185, | |
| "step": 2200, | |
| "valid_targets_mean": 5355.8, | |
| "valid_targets_min": 1673 | |
| }, | |
| { | |
| "epoch": 3.522364217252396, | |
| "grad_norm": 0.41147496462703653, | |
| "learning_rate": 2.327575937861612e-05, | |
| "loss": 0.1537, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13770106434822083, | |
| "step": 2205, | |
| "valid_targets_mean": 4609.2, | |
| "valid_targets_min": 1183 | |
| }, | |
| { | |
| "epoch": 3.5303514376996805, | |
| "grad_norm": 0.3934504837045112, | |
| "learning_rate": 2.3197134373413595e-05, | |
| "loss": 0.1443, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15065234899520874, | |
| "step": 2210, | |
| "valid_targets_mean": 5158.1, | |
| "valid_targets_min": 3030 | |
| }, | |
| { | |
| "epoch": 3.538338658146965, | |
| "grad_norm": 0.47040063543298133, | |
| "learning_rate": 2.3118458628672272e-05, | |
| "loss": 0.1592, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17348797619342804, | |
| "step": 2215, | |
| "valid_targets_mean": 4753.1, | |
| "valid_targets_min": 1514 | |
| }, | |
| { | |
| "epoch": 3.5463258785942493, | |
| "grad_norm": 0.35598059899413503, | |
| "learning_rate": 2.3039733393001233e-05, | |
| "loss": 0.1658, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13372495770454407, | |
| "step": 2220, | |
| "valid_targets_mean": 5608.9, | |
| "valid_targets_min": 2705 | |
| }, | |
| { | |
| "epoch": 3.5543130990415337, | |
| "grad_norm": 0.37488296056375664, | |
| "learning_rate": 2.2960959915794998e-05, | |
| "loss": 0.1498, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1291673332452774, | |
| "step": 2225, | |
| "valid_targets_mean": 5299.1, | |
| "valid_targets_min": 2487 | |
| }, | |
| { | |
| "epoch": 3.562300319488818, | |
| "grad_norm": 0.40529673808980154, | |
| "learning_rate": 2.2882139447213673e-05, | |
| "loss": 0.1393, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15351450443267822, | |
| "step": 2230, | |
| "valid_targets_mean": 4994.6, | |
| "valid_targets_min": 2786 | |
| }, | |
| { | |
| "epoch": 3.5702875399361025, | |
| "grad_norm": 0.40377627032667335, | |
| "learning_rate": 2.280327323816317e-05, | |
| "loss": 0.1435, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1551814079284668, | |
| "step": 2235, | |
| "valid_targets_mean": 5307.6, | |
| "valid_targets_min": 2958 | |
| }, | |
| { | |
| "epoch": 3.5782747603833864, | |
| "grad_norm": 0.3952507882872263, | |
| "learning_rate": 2.2724362540275288e-05, | |
| "loss": 0.1531, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14203189313411713, | |
| "step": 2240, | |
| "valid_targets_mean": 4896.2, | |
| "valid_targets_min": 2337 | |
| }, | |
| { | |
| "epoch": 3.586261980830671, | |
| "grad_norm": 0.40431897447692067, | |
| "learning_rate": 2.2645408605887882e-05, | |
| "loss": 0.1488, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13299547135829926, | |
| "step": 2245, | |
| "valid_targets_mean": 5133.1, | |
| "valid_targets_min": 2267 | |
| }, | |
| { | |
| "epoch": 3.594249201277955, | |
| "grad_norm": 0.40733331794258715, | |
| "learning_rate": 2.2566412688025003e-05, | |
| "loss": 0.1478, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14882946014404297, | |
| "step": 2250, | |
| "valid_targets_mean": 5328.9, | |
| "valid_targets_min": 2709 | |
| }, | |
| { | |
| "epoch": 3.6022364217252396, | |
| "grad_norm": 0.3713219175955481, | |
| "learning_rate": 2.2487376040376958e-05, | |
| "loss": 0.1528, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13275527954101562, | |
| "step": 2255, | |
| "valid_targets_mean": 5108.8, | |
| "valid_targets_min": 374 | |
| }, | |
| { | |
| "epoch": 3.610223642172524, | |
| "grad_norm": 0.393986516880644, | |
| "learning_rate": 2.240829991728048e-05, | |
| "loss": 0.1509, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15260398387908936, | |
| "step": 2260, | |
| "valid_targets_mean": 5226.6, | |
| "valid_targets_min": 3015 | |
| }, | |
| { | |
| "epoch": 3.6182108626198084, | |
| "grad_norm": 0.3626290958024588, | |
| "learning_rate": 2.2329185573698778e-05, | |
| "loss": 0.1476, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13706724345684052, | |
| "step": 2265, | |
| "valid_targets_mean": 5499.6, | |
| "valid_targets_min": 2477 | |
| }, | |
| { | |
| "epoch": 3.626198083067093, | |
| "grad_norm": 0.4623153256117741, | |
| "learning_rate": 2.225003426520163e-05, | |
| "loss": 0.1584, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16722369194030762, | |
| "step": 2270, | |
| "valid_targets_mean": 5215.1, | |
| "valid_targets_min": 2702 | |
| }, | |
| { | |
| "epoch": 3.6341853035143767, | |
| "grad_norm": 0.3814751686349694, | |
| "learning_rate": 2.2170847247945466e-05, | |
| "loss": 0.1619, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1622811257839203, | |
| "step": 2275, | |
| "valid_targets_mean": 5764.9, | |
| "valid_targets_min": 1571 | |
| }, | |
| { | |
| "epoch": 3.642172523961661, | |
| "grad_norm": 0.43631006700319885, | |
| "learning_rate": 2.209162577865342e-05, | |
| "loss": 0.1573, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16662268340587616, | |
| "step": 2280, | |
| "valid_targets_mean": 4647.3, | |
| "valid_targets_min": 969 | |
| }, | |
| { | |
| "epoch": 3.6501597444089455, | |
| "grad_norm": 0.432828912612411, | |
| "learning_rate": 2.201237111459539e-05, | |
| "loss": 0.149, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14495493471622467, | |
| "step": 2285, | |
| "valid_targets_mean": 4343.4, | |
| "valid_targets_min": 1515 | |
| }, | |
| { | |
| "epoch": 3.65814696485623, | |
| "grad_norm": 0.3963765136341493, | |
| "learning_rate": 2.193308451356809e-05, | |
| "loss": 0.1609, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1367836743593216, | |
| "step": 2290, | |
| "valid_targets_mean": 5635.3, | |
| "valid_targets_min": 3376 | |
| }, | |
| { | |
| "epoch": 3.6661341853035143, | |
| "grad_norm": 0.4072323437452567, | |
| "learning_rate": 2.1853767233875084e-05, | |
| "loss": 0.1547, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14956004917621613, | |
| "step": 2295, | |
| "valid_targets_mean": 5429.9, | |
| "valid_targets_min": 951 | |
| }, | |
| { | |
| "epoch": 3.6741214057507987, | |
| "grad_norm": 0.37136616042848897, | |
| "learning_rate": 2.1774420534306812e-05, | |
| "loss": 0.151, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1503937691450119, | |
| "step": 2300, | |
| "valid_targets_mean": 5513.6, | |
| "valid_targets_min": 3257 | |
| }, | |
| { | |
| "epoch": 3.682108626198083, | |
| "grad_norm": 0.39864550628099366, | |
| "learning_rate": 2.1695045674120615e-05, | |
| "loss": 0.1417, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14195352792739868, | |
| "step": 2305, | |
| "valid_targets_mean": 4920.6, | |
| "valid_targets_min": 1604 | |
| }, | |
| { | |
| "epoch": 3.6900958466453675, | |
| "grad_norm": 0.3914923357921744, | |
| "learning_rate": 2.1615643913020755e-05, | |
| "loss": 0.1492, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1457861065864563, | |
| "step": 2310, | |
| "valid_targets_mean": 5147.0, | |
| "valid_targets_min": 2466 | |
| }, | |
| { | |
| "epoch": 3.698083067092652, | |
| "grad_norm": 0.4411912188165477, | |
| "learning_rate": 2.1536216511138423e-05, | |
| "loss": 0.1424, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1481896936893463, | |
| "step": 2315, | |
| "valid_targets_mean": 3985.7, | |
| "valid_targets_min": 627 | |
| }, | |
| { | |
| "epoch": 3.7060702875399363, | |
| "grad_norm": 0.44973513726848635, | |
| "learning_rate": 2.145676472901174e-05, | |
| "loss": 0.1572, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15559715032577515, | |
| "step": 2320, | |
| "valid_targets_mean": 4583.7, | |
| "valid_targets_min": 2106 | |
| }, | |
| { | |
| "epoch": 3.7140575079872207, | |
| "grad_norm": 0.38060309801158404, | |
| "learning_rate": 2.1377289827565734e-05, | |
| "loss": 0.1518, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13940852880477905, | |
| "step": 2325, | |
| "valid_targets_mean": 4892.8, | |
| "valid_targets_min": 2134 | |
| }, | |
| { | |
| "epoch": 3.722044728434505, | |
| "grad_norm": 0.4318550804329095, | |
| "learning_rate": 2.129779306809236e-05, | |
| "loss": 0.1524, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1502557247877121, | |
| "step": 2330, | |
| "valid_targets_mean": 4569.1, | |
| "valid_targets_min": 275 | |
| }, | |
| { | |
| "epoch": 3.730031948881789, | |
| "grad_norm": 0.41004897426097164, | |
| "learning_rate": 2.121827571223046e-05, | |
| "loss": 0.1405, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14614850282669067, | |
| "step": 2335, | |
| "valid_targets_mean": 5097.9, | |
| "valid_targets_min": 1747 | |
| }, | |
| { | |
| "epoch": 3.7380191693290734, | |
| "grad_norm": 0.40299579830930937, | |
| "learning_rate": 2.113873902194575e-05, | |
| "loss": 0.153, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15635550022125244, | |
| "step": 2340, | |
| "valid_targets_mean": 5389.6, | |
| "valid_targets_min": 2928 | |
| }, | |
| { | |
| "epoch": 3.746006389776358, | |
| "grad_norm": 0.40881578446239625, | |
| "learning_rate": 2.1059184259510786e-05, | |
| "loss": 0.1312, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13039812445640564, | |
| "step": 2345, | |
| "valid_targets_mean": 4697.4, | |
| "valid_targets_min": 1512 | |
| }, | |
| { | |
| "epoch": 3.753993610223642, | |
| "grad_norm": 0.5081708478565543, | |
| "learning_rate": 2.0979612687484935e-05, | |
| "loss": 0.1381, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1455729901790619, | |
| "step": 2350, | |
| "valid_targets_mean": 5053.9, | |
| "valid_targets_min": 2862 | |
| }, | |
| { | |
| "epoch": 3.7619808306709266, | |
| "grad_norm": 0.3806214606104205, | |
| "learning_rate": 2.0900025568694334e-05, | |
| "loss": 0.1543, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.137091726064682, | |
| "step": 2355, | |
| "valid_targets_mean": 5183.1, | |
| "valid_targets_min": 2918 | |
| }, | |
| { | |
| "epoch": 3.769968051118211, | |
| "grad_norm": 0.4067074793960066, | |
| "learning_rate": 2.082042416621187e-05, | |
| "loss": 0.1611, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16555792093276978, | |
| "step": 2360, | |
| "valid_targets_mean": 5315.8, | |
| "valid_targets_min": 2396 | |
| }, | |
| { | |
| "epoch": 3.777955271565495, | |
| "grad_norm": 0.3843787513319819, | |
| "learning_rate": 2.0740809743337095e-05, | |
| "loss": 0.1532, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13774700462818146, | |
| "step": 2365, | |
| "valid_targets_mean": 5224.2, | |
| "valid_targets_min": 1909 | |
| }, | |
| { | |
| "epoch": 3.7859424920127793, | |
| "grad_norm": 0.4200093396326064, | |
| "learning_rate": 2.0661183563576202e-05, | |
| "loss": 0.1578, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.168784037232399, | |
| "step": 2370, | |
| "valid_targets_mean": 5614.2, | |
| "valid_targets_min": 1701 | |
| }, | |
| { | |
| "epoch": 3.7939297124600637, | |
| "grad_norm": 0.42594955249283983, | |
| "learning_rate": 2.0581546890621986e-05, | |
| "loss": 0.1527, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1369417905807495, | |
| "step": 2375, | |
| "valid_targets_mean": 4328.6, | |
| "valid_targets_min": 1427 | |
| }, | |
| { | |
| "epoch": 3.801916932907348, | |
| "grad_norm": 0.38816134242669587, | |
| "learning_rate": 2.0501900988333755e-05, | |
| "loss": 0.1506, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15396389365196228, | |
| "step": 2380, | |
| "valid_targets_mean": 5375.2, | |
| "valid_targets_min": 2013 | |
| }, | |
| { | |
| "epoch": 3.8099041533546325, | |
| "grad_norm": 0.435649647694377, | |
| "learning_rate": 2.0422247120717304e-05, | |
| "loss": 0.1487, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16302108764648438, | |
| "step": 2385, | |
| "valid_targets_mean": 4684.1, | |
| "valid_targets_min": 1956 | |
| }, | |
| { | |
| "epoch": 3.817891373801917, | |
| "grad_norm": 0.4251470592225304, | |
| "learning_rate": 2.0342586551904824e-05, | |
| "loss": 0.1492, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14779290556907654, | |
| "step": 2390, | |
| "valid_targets_mean": 4743.6, | |
| "valid_targets_min": 1118 | |
| }, | |
| { | |
| "epoch": 3.8258785942492013, | |
| "grad_norm": 0.36269855050925853, | |
| "learning_rate": 2.026292054613486e-05, | |
| "loss": 0.1449, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13070446252822876, | |
| "step": 2395, | |
| "valid_targets_mean": 5622.7, | |
| "valid_targets_min": 2033 | |
| }, | |
| { | |
| "epoch": 3.8338658146964857, | |
| "grad_norm": 0.4004742496039476, | |
| "learning_rate": 2.0183250367732253e-05, | |
| "loss": 0.1533, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15732410550117493, | |
| "step": 2400, | |
| "valid_targets_mean": 5364.6, | |
| "valid_targets_min": 2599 | |
| }, | |
| { | |
| "epoch": 3.84185303514377, | |
| "grad_norm": 0.4207703838080751, | |
| "learning_rate": 2.0103577281088065e-05, | |
| "loss": 0.1616, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12939991056919098, | |
| "step": 2405, | |
| "valid_targets_mean": 5015.5, | |
| "valid_targets_min": 3085 | |
| }, | |
| { | |
| "epoch": 3.8498402555910545, | |
| "grad_norm": 0.38774242069970005, | |
| "learning_rate": 2.00239025506395e-05, | |
| "loss": 0.1448, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14627763628959656, | |
| "step": 2410, | |
| "valid_targets_mean": 5791.6, | |
| "valid_targets_min": 2916 | |
| }, | |
| { | |
| "epoch": 3.857827476038339, | |
| "grad_norm": 0.411291074272304, | |
| "learning_rate": 1.9944227440849856e-05, | |
| "loss": 0.1485, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13100185990333557, | |
| "step": 2415, | |
| "valid_targets_mean": 4967.4, | |
| "valid_targets_min": 2574 | |
| }, | |
| { | |
| "epoch": 3.8658146964856233, | |
| "grad_norm": 0.4517275572750298, | |
| "learning_rate": 1.9864553216188454e-05, | |
| "loss": 0.1508, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16096553206443787, | |
| "step": 2420, | |
| "valid_targets_mean": 4204.8, | |
| "valid_targets_min": 1539 | |
| }, | |
| { | |
| "epoch": 3.873801916932907, | |
| "grad_norm": 0.4233673777607511, | |
| "learning_rate": 1.978488114111057e-05, | |
| "loss": 0.1468, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17223328351974487, | |
| "step": 2425, | |
| "valid_targets_mean": 4944.2, | |
| "valid_targets_min": 918 | |
| }, | |
| { | |
| "epoch": 3.8817891373801916, | |
| "grad_norm": 0.3828425432656369, | |
| "learning_rate": 1.970521248003735e-05, | |
| "loss": 0.15, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1449100375175476, | |
| "step": 2430, | |
| "valid_targets_mean": 5368.4, | |
| "valid_targets_min": 2409 | |
| }, | |
| { | |
| "epoch": 3.889776357827476, | |
| "grad_norm": 0.3734353013978433, | |
| "learning_rate": 1.9625548497335783e-05, | |
| "loss": 0.1618, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1504458487033844, | |
| "step": 2435, | |
| "valid_targets_mean": 6081.7, | |
| "valid_targets_min": 2614 | |
| }, | |
| { | |
| "epoch": 3.8977635782747604, | |
| "grad_norm": 0.4198890670596281, | |
| "learning_rate": 1.9545890457298592e-05, | |
| "loss": 0.1696, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18291476368904114, | |
| "step": 2440, | |
| "valid_targets_mean": 5823.9, | |
| "valid_targets_min": 1212 | |
| }, | |
| { | |
| "epoch": 3.905750798722045, | |
| "grad_norm": 0.4082169766432791, | |
| "learning_rate": 1.9466239624124203e-05, | |
| "loss": 0.1495, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16215559840202332, | |
| "step": 2445, | |
| "valid_targets_mean": 5081.9, | |
| "valid_targets_min": 2623 | |
| }, | |
| { | |
| "epoch": 3.913738019169329, | |
| "grad_norm": 0.42507759391580796, | |
| "learning_rate": 1.938659726189663e-05, | |
| "loss": 0.1664, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18141093850135803, | |
| "step": 2450, | |
| "valid_targets_mean": 5700.6, | |
| "valid_targets_min": 3488 | |
| }, | |
| { | |
| "epoch": 3.9217252396166136, | |
| "grad_norm": 0.386853712074672, | |
| "learning_rate": 1.9306964634565505e-05, | |
| "loss": 0.151, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.148773655295372, | |
| "step": 2455, | |
| "valid_targets_mean": 5347.9, | |
| "valid_targets_min": 1941 | |
| }, | |
| { | |
| "epoch": 3.9297124600638975, | |
| "grad_norm": 0.4286632380424892, | |
| "learning_rate": 1.9227343005925926e-05, | |
| "loss": 0.1613, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17419970035552979, | |
| "step": 2460, | |
| "valid_targets_mean": 4997.3, | |
| "valid_targets_min": 1964 | |
| }, | |
| { | |
| "epoch": 3.937699680511182, | |
| "grad_norm": 0.43509447170600585, | |
| "learning_rate": 1.914773363959845e-05, | |
| "loss": 0.148, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.138102188706398, | |
| "step": 2465, | |
| "valid_targets_mean": 4446.8, | |
| "valid_targets_min": 1158 | |
| }, | |
| { | |
| "epoch": 3.9456869009584663, | |
| "grad_norm": 0.40609683742403174, | |
| "learning_rate": 1.906813779900903e-05, | |
| "loss": 0.1566, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15796974301338196, | |
| "step": 2470, | |
| "valid_targets_mean": 5245.2, | |
| "valid_targets_min": 2569 | |
| }, | |
| { | |
| "epoch": 3.9536741214057507, | |
| "grad_norm": 0.418519387547944, | |
| "learning_rate": 1.8988556747368953e-05, | |
| "loss": 0.1504, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14734810590744019, | |
| "step": 2475, | |
| "valid_targets_mean": 4880.1, | |
| "valid_targets_min": 1891 | |
| }, | |
| { | |
| "epoch": 3.961661341853035, | |
| "grad_norm": 0.5463109840341152, | |
| "learning_rate": 1.890899174765481e-05, | |
| "loss": 0.1528, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14559370279312134, | |
| "step": 2480, | |
| "valid_targets_mean": 4150.4, | |
| "valid_targets_min": 1341 | |
| }, | |
| { | |
| "epoch": 3.9696485623003195, | |
| "grad_norm": 0.36832441479128725, | |
| "learning_rate": 1.8829444062588433e-05, | |
| "loss": 0.1475, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14017799496650696, | |
| "step": 2485, | |
| "valid_targets_mean": 5482.6, | |
| "valid_targets_min": 1872 | |
| }, | |
| { | |
| "epoch": 3.977635782747604, | |
| "grad_norm": 0.40833179950456444, | |
| "learning_rate": 1.8749914954616882e-05, | |
| "loss": 0.1502, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1394728124141693, | |
| "step": 2490, | |
| "valid_targets_mean": 4307.8, | |
| "valid_targets_min": 903 | |
| }, | |
| { | |
| "epoch": 3.9856230031948883, | |
| "grad_norm": 0.4026078671749823, | |
| "learning_rate": 1.867040568589236e-05, | |
| "loss": 0.161, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15371136367321014, | |
| "step": 2495, | |
| "valid_targets_mean": 4868.9, | |
| "valid_targets_min": 1699 | |
| }, | |
| { | |
| "epoch": 3.9936102236421727, | |
| "grad_norm": 0.4272683426872592, | |
| "learning_rate": 1.859091751825225e-05, | |
| "loss": 0.1617, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1574055254459381, | |
| "step": 2500, | |
| "valid_targets_mean": 4676.3, | |
| "valid_targets_min": 1277 | |
| }, | |
| { | |
| "epoch": 4.001597444089457, | |
| "grad_norm": 0.3966461226353622, | |
| "learning_rate": 1.8511451713199038e-05, | |
| "loss": 0.1583, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1441224217414856, | |
| "step": 2505, | |
| "valid_targets_mean": 5037.6, | |
| "valid_targets_min": 1868 | |
| }, | |
| { | |
| "epoch": 4.0095846645367414, | |
| "grad_norm": 0.39533317381527283, | |
| "learning_rate": 1.843200953188031e-05, | |
| "loss": 0.146, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1328619420528412, | |
| "step": 2510, | |
| "valid_targets_mean": 5424.0, | |
| "valid_targets_min": 2478 | |
| }, | |
| { | |
| "epoch": 4.017571884984026, | |
| "grad_norm": 0.47501216566879617, | |
| "learning_rate": 1.8352592235068735e-05, | |
| "loss": 0.1409, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14926929771900177, | |
| "step": 2515, | |
| "valid_targets_mean": 4750.3, | |
| "valid_targets_min": 720 | |
| }, | |
| { | |
| "epoch": 4.02555910543131, | |
| "grad_norm": 0.4431256435419237, | |
| "learning_rate": 1.8273201083142056e-05, | |
| "loss": 0.1412, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13658151030540466, | |
| "step": 2520, | |
| "valid_targets_mean": 5219.1, | |
| "valid_targets_min": 2240 | |
| }, | |
| { | |
| "epoch": 4.033546325878595, | |
| "grad_norm": 0.4341052169527362, | |
| "learning_rate": 1.8193837336063096e-05, | |
| "loss": 0.1355, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13655325770378113, | |
| "step": 2525, | |
| "valid_targets_mean": 4804.9, | |
| "valid_targets_min": 3148 | |
| }, | |
| { | |
| "epoch": 4.041533546325879, | |
| "grad_norm": 0.4274343667790037, | |
| "learning_rate": 1.8114502253359745e-05, | |
| "loss": 0.1406, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13518372178077698, | |
| "step": 2530, | |
| "valid_targets_mean": 4610.3, | |
| "valid_targets_min": 2005 | |
| }, | |
| { | |
| "epoch": 4.0495207667731625, | |
| "grad_norm": 0.3960597935382846, | |
| "learning_rate": 1.8035197094104986e-05, | |
| "loss": 0.1412, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14935217797756195, | |
| "step": 2535, | |
| "valid_targets_mean": 5497.9, | |
| "valid_targets_min": 1071 | |
| }, | |
| { | |
| "epoch": 4.057507987220447, | |
| "grad_norm": 0.3932279533626889, | |
| "learning_rate": 1.7955923116896908e-05, | |
| "loss": 0.1445, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13884808123111725, | |
| "step": 2540, | |
| "valid_targets_mean": 5324.1, | |
| "valid_targets_min": 2985 | |
| }, | |
| { | |
| "epoch": 4.065495207667731, | |
| "grad_norm": 0.44658920567704774, | |
| "learning_rate": 1.787668157983872e-05, | |
| "loss": 0.1349, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12284763157367706, | |
| "step": 2545, | |
| "valid_targets_mean": 5101.3, | |
| "valid_targets_min": 2557 | |
| }, | |
| { | |
| "epoch": 4.073482428115016, | |
| "grad_norm": 0.4363186848147753, | |
| "learning_rate": 1.779747374051881e-05, | |
| "loss": 0.1419, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1479886770248413, | |
| "step": 2550, | |
| "valid_targets_mean": 5137.4, | |
| "valid_targets_min": 2527 | |
| }, | |
| { | |
| "epoch": 4.0814696485623, | |
| "grad_norm": 0.38990260078321587, | |
| "learning_rate": 1.7718300855990767e-05, | |
| "loss": 0.1297, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1255975067615509, | |
| "step": 2555, | |
| "valid_targets_mean": 5238.6, | |
| "valid_targets_min": 1660 | |
| }, | |
| { | |
| "epoch": 4.0894568690095845, | |
| "grad_norm": 0.3829676714987014, | |
| "learning_rate": 1.7639164182753434e-05, | |
| "loss": 0.1468, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12952443957328796, | |
| "step": 2560, | |
| "valid_targets_mean": 5804.6, | |
| "valid_targets_min": 3038 | |
| }, | |
| { | |
| "epoch": 4.097444089456869, | |
| "grad_norm": 0.4467824407867215, | |
| "learning_rate": 1.7560064976730967e-05, | |
| "loss": 0.1458, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1782253533601761, | |
| "step": 2565, | |
| "valid_targets_mean": 5426.5, | |
| "valid_targets_min": 2708 | |
| }, | |
| { | |
| "epoch": 4.105431309904153, | |
| "grad_norm": 0.5615220450629616, | |
| "learning_rate": 1.7481004493252915e-05, | |
| "loss": 0.1512, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1406998187303543, | |
| "step": 2570, | |
| "valid_targets_mean": 5493.8, | |
| "valid_targets_min": 3307 | |
| }, | |
| { | |
| "epoch": 4.113418530351438, | |
| "grad_norm": 0.4074612668871005, | |
| "learning_rate": 1.740198398703428e-05, | |
| "loss": 0.1348, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11592158675193787, | |
| "step": 2575, | |
| "valid_targets_mean": 5014.1, | |
| "valid_targets_min": 1797 | |
| }, | |
| { | |
| "epoch": 4.121405750798722, | |
| "grad_norm": 0.40878424323849405, | |
| "learning_rate": 1.7323004712155613e-05, | |
| "loss": 0.1359, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14485692977905273, | |
| "step": 2580, | |
| "valid_targets_mean": 5438.9, | |
| "valid_targets_min": 2833 | |
| }, | |
| { | |
| "epoch": 4.1293929712460065, | |
| "grad_norm": 0.4495873466889326, | |
| "learning_rate": 1.724406792204312e-05, | |
| "loss": 0.1355, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12828205525875092, | |
| "step": 2585, | |
| "valid_targets_mean": 4479.4, | |
| "valid_targets_min": 903 | |
| }, | |
| { | |
| "epoch": 4.137380191693291, | |
| "grad_norm": 0.4253852992626158, | |
| "learning_rate": 1.7165174869448738e-05, | |
| "loss": 0.1376, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14189516007900238, | |
| "step": 2590, | |
| "valid_targets_mean": 5626.5, | |
| "valid_targets_min": 1959 | |
| }, | |
| { | |
| "epoch": 4.145367412140575, | |
| "grad_norm": 0.43748083888632144, | |
| "learning_rate": 1.7086326806430298e-05, | |
| "loss": 0.1336, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.135626882314682, | |
| "step": 2595, | |
| "valid_targets_mean": 5076.9, | |
| "valid_targets_min": 2455 | |
| }, | |
| { | |
| "epoch": 4.15335463258786, | |
| "grad_norm": 0.38131493213200895, | |
| "learning_rate": 1.7007524984331623e-05, | |
| "loss": 0.1518, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12402436137199402, | |
| "step": 2600, | |
| "valid_targets_mean": 6088.9, | |
| "valid_targets_min": 1073 | |
| }, | |
| { | |
| "epoch": 4.161341853035144, | |
| "grad_norm": 0.40616924571961455, | |
| "learning_rate": 1.692877065376268e-05, | |
| "loss": 0.139, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12958721816539764, | |
| "step": 2605, | |
| "valid_targets_mean": 5496.2, | |
| "valid_targets_min": 2810 | |
| }, | |
| { | |
| "epoch": 4.169329073482428, | |
| "grad_norm": 0.4124300193328953, | |
| "learning_rate": 1.6850065064579724e-05, | |
| "loss": 0.1301, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13094055652618408, | |
| "step": 2610, | |
| "valid_targets_mean": 4931.8, | |
| "valid_targets_min": 1008 | |
| }, | |
| { | |
| "epoch": 4.177316293929713, | |
| "grad_norm": 0.5432489590016187, | |
| "learning_rate": 1.677140946586548e-05, | |
| "loss": 0.1445, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1453172117471695, | |
| "step": 2615, | |
| "valid_targets_mean": 5214.3, | |
| "valid_targets_min": 1717 | |
| }, | |
| { | |
| "epoch": 4.185303514376997, | |
| "grad_norm": 0.4559868009421075, | |
| "learning_rate": 1.66928051059093e-05, | |
| "loss": 0.1399, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12931229174137115, | |
| "step": 2620, | |
| "valid_targets_mean": 4284.8, | |
| "valid_targets_min": 2350 | |
| }, | |
| { | |
| "epoch": 4.193290734824281, | |
| "grad_norm": 0.44499265896355733, | |
| "learning_rate": 1.6614253232187363e-05, | |
| "loss": 0.1496, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1416967213153839, | |
| "step": 2625, | |
| "valid_targets_mean": 4899.1, | |
| "valid_targets_min": 1878 | |
| }, | |
| { | |
| "epoch": 4.201277955271565, | |
| "grad_norm": 0.4448851827784073, | |
| "learning_rate": 1.6535755091342876e-05, | |
| "loss": 0.1359, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13339129090309143, | |
| "step": 2630, | |
| "valid_targets_mean": 4251.1, | |
| "valid_targets_min": 394 | |
| }, | |
| { | |
| "epoch": 4.2092651757188495, | |
| "grad_norm": 0.4223973968424472, | |
| "learning_rate": 1.6457311929166286e-05, | |
| "loss": 0.1401, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1536128968000412, | |
| "step": 2635, | |
| "valid_targets_mean": 5103.8, | |
| "valid_targets_min": 2616 | |
| }, | |
| { | |
| "epoch": 4.217252396166134, | |
| "grad_norm": 0.41075331289309785, | |
| "learning_rate": 1.637892499057551e-05, | |
| "loss": 0.1434, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13937146961688995, | |
| "step": 2640, | |
| "valid_targets_mean": 5179.8, | |
| "valid_targets_min": 1636 | |
| }, | |
| { | |
| "epoch": 4.225239616613418, | |
| "grad_norm": 0.5234615024279543, | |
| "learning_rate": 1.6300595519596177e-05, | |
| "loss": 0.1379, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18650765717029572, | |
| "step": 2645, | |
| "valid_targets_mean": 5131.8, | |
| "valid_targets_min": 1585 | |
| }, | |
| { | |
| "epoch": 4.233226837060703, | |
| "grad_norm": 0.48457019057640444, | |
| "learning_rate": 1.6222324759341897e-05, | |
| "loss": 0.1424, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14671725034713745, | |
| "step": 2650, | |
| "valid_targets_mean": 4458.0, | |
| "valid_targets_min": 300 | |
| }, | |
| { | |
| "epoch": 4.241214057507987, | |
| "grad_norm": 0.42614781653645123, | |
| "learning_rate": 1.6144113951994515e-05, | |
| "loss": 0.1494, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13661018013954163, | |
| "step": 2655, | |
| "valid_targets_mean": 4837.0, | |
| "valid_targets_min": 2120 | |
| }, | |
| { | |
| "epoch": 4.2492012779552715, | |
| "grad_norm": 0.44209702643251814, | |
| "learning_rate": 1.60659643387844e-05, | |
| "loss": 0.1462, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13152384757995605, | |
| "step": 2660, | |
| "valid_targets_mean": 4384.4, | |
| "valid_targets_min": 1978 | |
| }, | |
| { | |
| "epoch": 4.257188498402556, | |
| "grad_norm": 0.39830283502498515, | |
| "learning_rate": 1.5987877159970765e-05, | |
| "loss": 0.1419, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12998461723327637, | |
| "step": 2665, | |
| "valid_targets_mean": 5367.4, | |
| "valid_targets_min": 2955 | |
| }, | |
| { | |
| "epoch": 4.26517571884984, | |
| "grad_norm": 0.4480130200739297, | |
| "learning_rate": 1.5909853654821947e-05, | |
| "loss": 0.1537, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15372559428215027, | |
| "step": 2670, | |
| "valid_targets_mean": 5901.6, | |
| "valid_targets_min": 3244 | |
| }, | |
| { | |
| "epoch": 4.273162939297125, | |
| "grad_norm": 0.4493912895490064, | |
| "learning_rate": 1.583189506159579e-05, | |
| "loss": 0.1384, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16665223240852356, | |
| "step": 2675, | |
| "valid_targets_mean": 4816.3, | |
| "valid_targets_min": 1771 | |
| }, | |
| { | |
| "epoch": 4.281150159744409, | |
| "grad_norm": 0.3775039063894273, | |
| "learning_rate": 1.575400261751993e-05, | |
| "loss": 0.1314, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12325578927993774, | |
| "step": 2680, | |
| "valid_targets_mean": 5259.9, | |
| "valid_targets_min": 2060 | |
| }, | |
| { | |
| "epoch": 4.289137380191693, | |
| "grad_norm": 0.36242748304539807, | |
| "learning_rate": 1.567617755877223e-05, | |
| "loss": 0.1326, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1113056093454361, | |
| "step": 2685, | |
| "valid_targets_mean": 5679.1, | |
| "valid_targets_min": 2267 | |
| }, | |
| { | |
| "epoch": 4.297124600638978, | |
| "grad_norm": 0.3782399320474647, | |
| "learning_rate": 1.5598421120461106e-05, | |
| "loss": 0.1336, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13300839066505432, | |
| "step": 2690, | |
| "valid_targets_mean": 5625.7, | |
| "valid_targets_min": 2803 | |
| }, | |
| { | |
| "epoch": 4.305111821086262, | |
| "grad_norm": 0.4145284567557428, | |
| "learning_rate": 1.552073453660595e-05, | |
| "loss": 0.1402, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14893612265586853, | |
| "step": 2695, | |
| "valid_targets_mean": 5686.5, | |
| "valid_targets_min": 2771 | |
| }, | |
| { | |
| "epoch": 4.313099041533547, | |
| "grad_norm": 0.4807929597553624, | |
| "learning_rate": 1.5443119040117544e-05, | |
| "loss": 0.147, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15257103741168976, | |
| "step": 2700, | |
| "valid_targets_mean": 4542.7, | |
| "valid_targets_min": 1124 | |
| }, | |
| { | |
| "epoch": 4.321086261980831, | |
| "grad_norm": 0.4479697909421426, | |
| "learning_rate": 1.536557586277849e-05, | |
| "loss": 0.1488, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16626892983913422, | |
| "step": 2705, | |
| "valid_targets_mean": 4994.2, | |
| "valid_targets_min": 2836 | |
| }, | |
| { | |
| "epoch": 4.329073482428115, | |
| "grad_norm": 0.43481260876027245, | |
| "learning_rate": 1.5288106235223655e-05, | |
| "loss": 0.1334, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11133088916540146, | |
| "step": 2710, | |
| "valid_targets_mean": 5284.4, | |
| "valid_targets_min": 2343 | |
| }, | |
| { | |
| "epoch": 4.3370607028754, | |
| "grad_norm": 0.4414899149952816, | |
| "learning_rate": 1.521071138692066e-05, | |
| "loss": 0.1308, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13162463903427124, | |
| "step": 2715, | |
| "valid_targets_mean": 4788.9, | |
| "valid_targets_min": 2138 | |
| }, | |
| { | |
| "epoch": 4.345047923322683, | |
| "grad_norm": 0.43449123806206597, | |
| "learning_rate": 1.5133392546150347e-05, | |
| "loss": 0.14, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1456822156906128, | |
| "step": 2720, | |
| "valid_targets_mean": 4905.4, | |
| "valid_targets_min": 2887 | |
| }, | |
| { | |
| "epoch": 4.353035143769968, | |
| "grad_norm": 0.40224781938061027, | |
| "learning_rate": 1.5056150939987303e-05, | |
| "loss": 0.1466, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1567641645669937, | |
| "step": 2725, | |
| "valid_targets_mean": 5639.9, | |
| "valid_targets_min": 2727 | |
| }, | |
| { | |
| "epoch": 4.361022364217252, | |
| "grad_norm": 0.413264337443506, | |
| "learning_rate": 1.4978987794280363e-05, | |
| "loss": 0.1408, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14558567106723785, | |
| "step": 2730, | |
| "valid_targets_mean": 5333.1, | |
| "valid_targets_min": 2730 | |
| }, | |
| { | |
| "epoch": 4.3690095846645365, | |
| "grad_norm": 0.387834064984216, | |
| "learning_rate": 1.4901904333633179e-05, | |
| "loss": 0.1464, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1191043108701706, | |
| "step": 2735, | |
| "valid_targets_mean": 5340.6, | |
| "valid_targets_min": 839 | |
| }, | |
| { | |
| "epoch": 4.376996805111821, | |
| "grad_norm": 0.42328714214886426, | |
| "learning_rate": 1.4824901781384777e-05, | |
| "loss": 0.1451, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12494292855262756, | |
| "step": 2740, | |
| "valid_targets_mean": 5207.4, | |
| "valid_targets_min": 873 | |
| }, | |
| { | |
| "epoch": 4.384984025559105, | |
| "grad_norm": 0.41291641496627723, | |
| "learning_rate": 1.4747981359590138e-05, | |
| "loss": 0.1406, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14373239874839783, | |
| "step": 2745, | |
| "valid_targets_mean": 5233.2, | |
| "valid_targets_min": 620 | |
| }, | |
| { | |
| "epoch": 4.39297124600639, | |
| "grad_norm": 0.41680588869615315, | |
| "learning_rate": 1.4671144289000806e-05, | |
| "loss": 0.1467, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13409414887428284, | |
| "step": 2750, | |
| "valid_targets_mean": 4895.8, | |
| "valid_targets_min": 718 | |
| }, | |
| { | |
| "epoch": 4.400958466453674, | |
| "grad_norm": 0.43564033533599, | |
| "learning_rate": 1.4594391789045513e-05, | |
| "loss": 0.1336, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14286726713180542, | |
| "step": 2755, | |
| "valid_targets_mean": 4944.8, | |
| "valid_targets_min": 2580 | |
| }, | |
| { | |
| "epoch": 4.4089456869009584, | |
| "grad_norm": 0.4312838639423176, | |
| "learning_rate": 1.4517725077810835e-05, | |
| "loss": 0.14, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12913718819618225, | |
| "step": 2760, | |
| "valid_targets_mean": 4614.1, | |
| "valid_targets_min": 2618 | |
| }, | |
| { | |
| "epoch": 4.416932907348243, | |
| "grad_norm": 0.4656898451591246, | |
| "learning_rate": 1.4441145372021847e-05, | |
| "loss": 0.1449, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1477980762720108, | |
| "step": 2765, | |
| "valid_targets_mean": 4273.1, | |
| "valid_targets_min": 582 | |
| }, | |
| { | |
| "epoch": 4.424920127795527, | |
| "grad_norm": 0.41567112165116316, | |
| "learning_rate": 1.4364653887022831e-05, | |
| "loss": 0.143, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1353229582309723, | |
| "step": 2770, | |
| "valid_targets_mean": 5309.9, | |
| "valid_targets_min": 2005 | |
| }, | |
| { | |
| "epoch": 4.432907348242812, | |
| "grad_norm": 0.38949895638993, | |
| "learning_rate": 1.4288251836757956e-05, | |
| "loss": 0.1485, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11882094293832779, | |
| "step": 2775, | |
| "valid_targets_mean": 4802.4, | |
| "valid_targets_min": 1867 | |
| }, | |
| { | |
| "epoch": 4.440894568690096, | |
| "grad_norm": 0.4413617607164514, | |
| "learning_rate": 1.4211940433752052e-05, | |
| "loss": 0.1392, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15152302384376526, | |
| "step": 2780, | |
| "valid_targets_mean": 4913.9, | |
| "valid_targets_min": 1909 | |
| }, | |
| { | |
| "epoch": 4.44888178913738, | |
| "grad_norm": 0.45221967982638533, | |
| "learning_rate": 1.4135720889091342e-05, | |
| "loss": 0.1454, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14447245001792908, | |
| "step": 2785, | |
| "valid_targets_mean": 4912.3, | |
| "valid_targets_min": 1224 | |
| }, | |
| { | |
| "epoch": 4.456869009584665, | |
| "grad_norm": 0.43047197776782525, | |
| "learning_rate": 1.4059594412404227e-05, | |
| "loss": 0.1357, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14408227801322937, | |
| "step": 2790, | |
| "valid_targets_mean": 4614.1, | |
| "valid_targets_min": 2445 | |
| }, | |
| { | |
| "epoch": 4.464856230031949, | |
| "grad_norm": 0.3842880102075406, | |
| "learning_rate": 1.3983562211842094e-05, | |
| "loss": 0.1472, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13477754592895508, | |
| "step": 2795, | |
| "valid_targets_mean": 6273.9, | |
| "valid_targets_min": 2646 | |
| }, | |
| { | |
| "epoch": 4.472843450479234, | |
| "grad_norm": 0.4450220606318698, | |
| "learning_rate": 1.390762549406013e-05, | |
| "loss": 0.1388, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13454917073249817, | |
| "step": 2800, | |
| "valid_targets_mean": 4816.1, | |
| "valid_targets_min": 1420 | |
| }, | |
| { | |
| "epoch": 4.480830670926517, | |
| "grad_norm": 0.4529054789954243, | |
| "learning_rate": 1.3831785464198188e-05, | |
| "loss": 0.14, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1483660638332367, | |
| "step": 2805, | |
| "valid_targets_mean": 4292.0, | |
| "valid_targets_min": 1770 | |
| }, | |
| { | |
| "epoch": 4.488817891373802, | |
| "grad_norm": 0.45493579487070407, | |
| "learning_rate": 1.3756043325861648e-05, | |
| "loss": 0.1374, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13628098368644714, | |
| "step": 2810, | |
| "valid_targets_mean": 4936.7, | |
| "valid_targets_min": 721 | |
| }, | |
| { | |
| "epoch": 4.496805111821086, | |
| "grad_norm": 0.4045078196344025, | |
| "learning_rate": 1.3680400281102326e-05, | |
| "loss": 0.1464, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13625001907348633, | |
| "step": 2815, | |
| "valid_targets_mean": 5333.7, | |
| "valid_targets_min": 355 | |
| }, | |
| { | |
| "epoch": 4.50479233226837, | |
| "grad_norm": 0.4687460568836481, | |
| "learning_rate": 1.3604857530399378e-05, | |
| "loss": 0.1309, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14359992742538452, | |
| "step": 2820, | |
| "valid_targets_mean": 4495.5, | |
| "valid_targets_min": 1894 | |
| }, | |
| { | |
| "epoch": 4.512779552715655, | |
| "grad_norm": 0.42073419746521995, | |
| "learning_rate": 1.3529416272640278e-05, | |
| "loss": 0.1458, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13775482773780823, | |
| "step": 2825, | |
| "valid_targets_mean": 4943.0, | |
| "valid_targets_min": 437 | |
| }, | |
| { | |
| "epoch": 4.520766773162939, | |
| "grad_norm": 0.3775221900507281, | |
| "learning_rate": 1.3454077705101766e-05, | |
| "loss": 0.136, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10600955784320831, | |
| "step": 2830, | |
| "valid_targets_mean": 5095.6, | |
| "valid_targets_min": 3027 | |
| }, | |
| { | |
| "epoch": 4.5287539936102235, | |
| "grad_norm": 0.42611076885649124, | |
| "learning_rate": 1.3378843023430856e-05, | |
| "loss": 0.1386, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1378186047077179, | |
| "step": 2835, | |
| "valid_targets_mean": 4786.4, | |
| "valid_targets_min": 2178 | |
| }, | |
| { | |
| "epoch": 4.536741214057508, | |
| "grad_norm": 0.4006035055589416, | |
| "learning_rate": 1.3303713421625864e-05, | |
| "loss": 0.146, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14511029422283173, | |
| "step": 2840, | |
| "valid_targets_mean": 5419.2, | |
| "valid_targets_min": 3010 | |
| }, | |
| { | |
| "epoch": 4.544728434504792, | |
| "grad_norm": 0.4907357609828477, | |
| "learning_rate": 1.3228690092017452e-05, | |
| "loss": 0.1386, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16007882356643677, | |
| "step": 2845, | |
| "valid_targets_mean": 4437.4, | |
| "valid_targets_min": 2405 | |
| }, | |
| { | |
| "epoch": 4.552715654952077, | |
| "grad_norm": 0.41600679703088017, | |
| "learning_rate": 1.3153774225249705e-05, | |
| "loss": 0.1491, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14039240777492523, | |
| "step": 2850, | |
| "valid_targets_mean": 5159.4, | |
| "valid_targets_min": 3496 | |
| }, | |
| { | |
| "epoch": 4.560702875399361, | |
| "grad_norm": 0.3850308146165297, | |
| "learning_rate": 1.3078967010261243e-05, | |
| "loss": 0.1324, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12653343379497528, | |
| "step": 2855, | |
| "valid_targets_mean": 5327.7, | |
| "valid_targets_min": 2710 | |
| }, | |
| { | |
| "epoch": 4.568690095846645, | |
| "grad_norm": 0.4432256328948607, | |
| "learning_rate": 1.3004269634266357e-05, | |
| "loss": 0.1399, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1672753393650055, | |
| "step": 2860, | |
| "valid_targets_mean": 4823.9, | |
| "valid_targets_min": 1860 | |
| }, | |
| { | |
| "epoch": 4.57667731629393, | |
| "grad_norm": 0.4105601911005862, | |
| "learning_rate": 1.2929683282736135e-05, | |
| "loss": 0.13, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13038793206214905, | |
| "step": 2865, | |
| "valid_targets_mean": 5249.5, | |
| "valid_targets_min": 2195 | |
| }, | |
| { | |
| "epoch": 4.584664536741214, | |
| "grad_norm": 0.4498928388359759, | |
| "learning_rate": 1.285520913937969e-05, | |
| "loss": 0.1445, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1397016942501068, | |
| "step": 2870, | |
| "valid_targets_mean": 5130.2, | |
| "valid_targets_min": 2150 | |
| }, | |
| { | |
| "epoch": 4.592651757188499, | |
| "grad_norm": 0.3727764906898393, | |
| "learning_rate": 1.2780848386125354e-05, | |
| "loss": 0.1414, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11744374781847, | |
| "step": 2875, | |
| "valid_targets_mean": 5217.1, | |
| "valid_targets_min": 1796 | |
| }, | |
| { | |
| "epoch": 4.600638977635783, | |
| "grad_norm": 0.4509372704427557, | |
| "learning_rate": 1.2706602203101913e-05, | |
| "loss": 0.1477, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15393175184726715, | |
| "step": 2880, | |
| "valid_targets_mean": 4951.0, | |
| "valid_targets_min": 2528 | |
| }, | |
| { | |
| "epoch": 4.608626198083067, | |
| "grad_norm": 0.46472468950214796, | |
| "learning_rate": 1.2632471768619894e-05, | |
| "loss": 0.1326, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12830111384391785, | |
| "step": 2885, | |
| "valid_targets_mean": 5557.1, | |
| "valid_targets_min": 2607 | |
| }, | |
| { | |
| "epoch": 4.616613418530352, | |
| "grad_norm": 0.3830550928407993, | |
| "learning_rate": 1.255845825915286e-05, | |
| "loss": 0.1449, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14168532192707062, | |
| "step": 2890, | |
| "valid_targets_mean": 5933.2, | |
| "valid_targets_min": 2958 | |
| }, | |
| { | |
| "epoch": 4.624600638977636, | |
| "grad_norm": 0.4047988882634234, | |
| "learning_rate": 1.248456284931873e-05, | |
| "loss": 0.1369, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12535330653190613, | |
| "step": 2895, | |
| "valid_targets_mean": 5011.2, | |
| "valid_targets_min": 2884 | |
| }, | |
| { | |
| "epoch": 4.63258785942492, | |
| "grad_norm": 0.4343114279834168, | |
| "learning_rate": 1.241078671186115e-05, | |
| "loss": 0.1345, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15251968801021576, | |
| "step": 2900, | |
| "valid_targets_mean": 5684.7, | |
| "valid_targets_min": 2050 | |
| }, | |
| { | |
| "epoch": 4.640575079872205, | |
| "grad_norm": 0.42734389272030726, | |
| "learning_rate": 1.2337131017630873e-05, | |
| "loss": 0.1475, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1396963745355606, | |
| "step": 2905, | |
| "valid_targets_mean": 4954.3, | |
| "valid_targets_min": 2121 | |
| }, | |
| { | |
| "epoch": 4.6485623003194885, | |
| "grad_norm": 0.42562590504861497, | |
| "learning_rate": 1.2263596935567175e-05, | |
| "loss": 0.1568, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13401544094085693, | |
| "step": 2910, | |
| "valid_targets_mean": 4768.8, | |
| "valid_targets_min": 2378 | |
| }, | |
| { | |
| "epoch": 4.656549520766773, | |
| "grad_norm": 0.43311933495604993, | |
| "learning_rate": 1.2190185632679316e-05, | |
| "loss": 0.1429, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14377030730247498, | |
| "step": 2915, | |
| "valid_targets_mean": 4817.3, | |
| "valid_targets_min": 2440 | |
| }, | |
| { | |
| "epoch": 4.664536741214057, | |
| "grad_norm": 0.49944494427868796, | |
| "learning_rate": 1.2116898274028008e-05, | |
| "loss": 0.1396, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13626545667648315, | |
| "step": 2920, | |
| "valid_targets_mean": 5139.2, | |
| "valid_targets_min": 2775 | |
| }, | |
| { | |
| "epoch": 4.672523961661342, | |
| "grad_norm": 0.39926248446006474, | |
| "learning_rate": 1.2043736022706934e-05, | |
| "loss": 0.1379, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13161957263946533, | |
| "step": 2925, | |
| "valid_targets_mean": 5849.1, | |
| "valid_targets_min": 3651 | |
| }, | |
| { | |
| "epoch": 4.680511182108626, | |
| "grad_norm": 0.7922927611236957, | |
| "learning_rate": 1.1970700039824271e-05, | |
| "loss": 0.1469, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15535059571266174, | |
| "step": 2930, | |
| "valid_targets_mean": 5403.1, | |
| "valid_targets_min": 867 | |
| }, | |
| { | |
| "epoch": 4.68849840255591, | |
| "grad_norm": 0.4386234824779697, | |
| "learning_rate": 1.1897791484484284e-05, | |
| "loss": 0.1445, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15392088890075684, | |
| "step": 2935, | |
| "valid_targets_mean": 4912.0, | |
| "valid_targets_min": 1925 | |
| }, | |
| { | |
| "epoch": 4.696485623003195, | |
| "grad_norm": 0.4258719687350311, | |
| "learning_rate": 1.182501151376893e-05, | |
| "loss": 0.1408, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1344785988330841, | |
| "step": 2940, | |
| "valid_targets_mean": 5154.1, | |
| "valid_targets_min": 2529 | |
| }, | |
| { | |
| "epoch": 4.704472843450479, | |
| "grad_norm": 0.4228787447144483, | |
| "learning_rate": 1.1752361282719472e-05, | |
| "loss": 0.1417, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14930549263954163, | |
| "step": 2945, | |
| "valid_targets_mean": 5401.6, | |
| "valid_targets_min": 648 | |
| }, | |
| { | |
| "epoch": 4.712460063897764, | |
| "grad_norm": 0.4383411832453612, | |
| "learning_rate": 1.1679841944318171e-05, | |
| "loss": 0.1478, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17733189463615417, | |
| "step": 2950, | |
| "valid_targets_mean": 5753.0, | |
| "valid_targets_min": 2307 | |
| }, | |
| { | |
| "epoch": 4.720447284345048, | |
| "grad_norm": 0.4344229824281209, | |
| "learning_rate": 1.1607454649469991e-05, | |
| "loss": 0.1417, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13039609789848328, | |
| "step": 2955, | |
| "valid_targets_mean": 4847.1, | |
| "valid_targets_min": 2291 | |
| }, | |
| { | |
| "epoch": 4.728434504792332, | |
| "grad_norm": 0.42881475579507555, | |
| "learning_rate": 1.1535200546984291e-05, | |
| "loss": 0.1352, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14639699459075928, | |
| "step": 2960, | |
| "valid_targets_mean": 4676.6, | |
| "valid_targets_min": 973 | |
| }, | |
| { | |
| "epoch": 4.736421725239617, | |
| "grad_norm": 0.3784938281294308, | |
| "learning_rate": 1.1463080783556663e-05, | |
| "loss": 0.1392, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13301125168800354, | |
| "step": 2965, | |
| "valid_targets_mean": 5802.6, | |
| "valid_targets_min": 3165 | |
| }, | |
| { | |
| "epoch": 4.744408945686901, | |
| "grad_norm": 0.4339722135550065, | |
| "learning_rate": 1.139109650375066e-05, | |
| "loss": 0.131, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13034439086914062, | |
| "step": 2970, | |
| "valid_targets_mean": 4896.6, | |
| "valid_targets_min": 1280 | |
| }, | |
| { | |
| "epoch": 4.752396166134186, | |
| "grad_norm": 0.4084301632324553, | |
| "learning_rate": 1.1319248849979705e-05, | |
| "loss": 0.1443, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1479320228099823, | |
| "step": 2975, | |
| "valid_targets_mean": 5355.2, | |
| "valid_targets_min": 3137 | |
| }, | |
| { | |
| "epoch": 4.76038338658147, | |
| "grad_norm": 0.4777482882723074, | |
| "learning_rate": 1.1247538962488882e-05, | |
| "loss": 0.1348, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14577484130859375, | |
| "step": 2980, | |
| "valid_targets_mean": 4488.1, | |
| "valid_targets_min": 2280 | |
| }, | |
| { | |
| "epoch": 4.768370607028754, | |
| "grad_norm": 0.4392945434431228, | |
| "learning_rate": 1.1175967979336913e-05, | |
| "loss": 0.1463, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16310063004493713, | |
| "step": 2985, | |
| "valid_targets_mean": 4647.6, | |
| "valid_targets_min": 967 | |
| }, | |
| { | |
| "epoch": 4.776357827476039, | |
| "grad_norm": 0.4146186593725022, | |
| "learning_rate": 1.1104537036378054e-05, | |
| "loss": 0.1481, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13050082325935364, | |
| "step": 2990, | |
| "valid_targets_mean": 5022.1, | |
| "valid_targets_min": 2579 | |
| }, | |
| { | |
| "epoch": 4.784345047923322, | |
| "grad_norm": 0.42525737572914857, | |
| "learning_rate": 1.1033247267244063e-05, | |
| "loss": 0.1453, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15056374669075012, | |
| "step": 2995, | |
| "valid_targets_mean": 4960.4, | |
| "valid_targets_min": 555 | |
| }, | |
| { | |
| "epoch": 4.792332268370607, | |
| "grad_norm": 0.4199963701369293, | |
| "learning_rate": 1.0962099803326257e-05, | |
| "loss": 0.1393, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15005168318748474, | |
| "step": 3000, | |
| "valid_targets_mean": 5291.2, | |
| "valid_targets_min": 2743 | |
| }, | |
| { | |
| "epoch": 4.800319488817891, | |
| "grad_norm": 0.39998270208358677, | |
| "learning_rate": 1.089109577375748e-05, | |
| "loss": 0.1383, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13126075267791748, | |
| "step": 3005, | |
| "valid_targets_mean": 5130.5, | |
| "valid_targets_min": 907 | |
| }, | |
| { | |
| "epoch": 4.8083067092651754, | |
| "grad_norm": 0.442918402342634, | |
| "learning_rate": 1.082023630539427e-05, | |
| "loss": 0.1407, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16068477928638458, | |
| "step": 3010, | |
| "valid_targets_mean": 4675.6, | |
| "valid_targets_min": 2001 | |
| }, | |
| { | |
| "epoch": 4.81629392971246, | |
| "grad_norm": 0.41840288982461543, | |
| "learning_rate": 1.07495225227989e-05, | |
| "loss": 0.1464, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13850602507591248, | |
| "step": 3015, | |
| "valid_targets_mean": 4782.4, | |
| "valid_targets_min": 597 | |
| }, | |
| { | |
| "epoch": 4.824281150159744, | |
| "grad_norm": 0.4372393493558505, | |
| "learning_rate": 1.0678955548221595e-05, | |
| "loss": 0.1387, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14146071672439575, | |
| "step": 3020, | |
| "valid_targets_mean": 4743.4, | |
| "valid_targets_min": 2722 | |
| }, | |
| { | |
| "epoch": 4.832268370607029, | |
| "grad_norm": 0.3973596489807888, | |
| "learning_rate": 1.0608536501582654e-05, | |
| "loss": 0.135, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13409081101417542, | |
| "step": 3025, | |
| "valid_targets_mean": 5319.2, | |
| "valid_targets_min": 3129 | |
| }, | |
| { | |
| "epoch": 4.840255591054313, | |
| "grad_norm": 0.39467379072161, | |
| "learning_rate": 1.0538266500454739e-05, | |
| "loss": 0.1321, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11430096626281738, | |
| "step": 3030, | |
| "valid_targets_mean": 4836.1, | |
| "valid_targets_min": 430 | |
| }, | |
| { | |
| "epoch": 4.848242811501597, | |
| "grad_norm": 0.432567881914055, | |
| "learning_rate": 1.0468146660045118e-05, | |
| "loss": 0.1361, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14244410395622253, | |
| "step": 3035, | |
| "valid_targets_mean": 4793.7, | |
| "valid_targets_min": 1957 | |
| }, | |
| { | |
| "epoch": 4.856230031948882, | |
| "grad_norm": 0.4246076339273224, | |
| "learning_rate": 1.0398178093177928e-05, | |
| "loss": 0.1287, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12329506129026413, | |
| "step": 3040, | |
| "valid_targets_mean": 4797.9, | |
| "valid_targets_min": 2162 | |
| }, | |
| { | |
| "epoch": 4.864217252396166, | |
| "grad_norm": 0.4204609528026509, | |
| "learning_rate": 1.0328361910276592e-05, | |
| "loss": 0.1297, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13250909745693207, | |
| "step": 3045, | |
| "valid_targets_mean": 5006.5, | |
| "valid_targets_min": 1212 | |
| }, | |
| { | |
| "epoch": 4.872204472843451, | |
| "grad_norm": 0.425870763703628, | |
| "learning_rate": 1.0258699219346091e-05, | |
| "loss": 0.1422, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13689568638801575, | |
| "step": 3050, | |
| "valid_targets_mean": 5219.4, | |
| "valid_targets_min": 2995 | |
| }, | |
| { | |
| "epoch": 4.880191693290735, | |
| "grad_norm": 0.4682398817566339, | |
| "learning_rate": 1.0189191125955503e-05, | |
| "loss": 0.1419, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1536734402179718, | |
| "step": 3055, | |
| "valid_targets_mean": 4703.2, | |
| "valid_targets_min": 1916 | |
| }, | |
| { | |
| "epoch": 4.888178913738019, | |
| "grad_norm": 0.392890766624362, | |
| "learning_rate": 1.0119838733220342e-05, | |
| "loss": 0.1333, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11669045686721802, | |
| "step": 3060, | |
| "valid_targets_mean": 4893.2, | |
| "valid_targets_min": 2197 | |
| }, | |
| { | |
| "epoch": 4.896166134185304, | |
| "grad_norm": 0.4055242077223966, | |
| "learning_rate": 1.0050643141785148e-05, | |
| "loss": 0.1354, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12273713946342468, | |
| "step": 3065, | |
| "valid_targets_mean": 4444.8, | |
| "valid_targets_min": 276 | |
| }, | |
| { | |
| "epoch": 4.904153354632588, | |
| "grad_norm": 0.39303444400781423, | |
| "learning_rate": 9.981605449805933e-06, | |
| "loss": 0.1439, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.137748584151268, | |
| "step": 3070, | |
| "valid_targets_mean": 5663.9, | |
| "valid_targets_min": 2900 | |
| }, | |
| { | |
| "epoch": 4.912140575079873, | |
| "grad_norm": 0.3982552241100896, | |
| "learning_rate": 9.912726752932823e-06, | |
| "loss": 0.14, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1523684710264206, | |
| "step": 3075, | |
| "valid_targets_mean": 6004.7, | |
| "valid_targets_min": 2554 | |
| }, | |
| { | |
| "epoch": 4.920127795527156, | |
| "grad_norm": 0.5237594198391005, | |
| "learning_rate": 9.844008144292643e-06, | |
| "loss": 0.13, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13208919763565063, | |
| "step": 3080, | |
| "valid_targets_mean": 4878.6, | |
| "valid_targets_min": 1276 | |
| }, | |
| { | |
| "epoch": 4.928115015974441, | |
| "grad_norm": 0.41548966714713237, | |
| "learning_rate": 9.775450714471537e-06, | |
| "loss": 0.1517, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14208267629146576, | |
| "step": 3085, | |
| "valid_targets_mean": 5072.9, | |
| "valid_targets_min": 2925 | |
| }, | |
| { | |
| "epoch": 4.936102236421725, | |
| "grad_norm": 0.3786764132788114, | |
| "learning_rate": 9.707055551497734e-06, | |
| "loss": 0.1353, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1280408501625061, | |
| "step": 3090, | |
| "valid_targets_mean": 5729.2, | |
| "valid_targets_min": 3294 | |
| }, | |
| { | |
| "epoch": 4.944089456869009, | |
| "grad_norm": 0.42148209593047986, | |
| "learning_rate": 9.638823740824177e-06, | |
| "loss": 0.126, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1342983841896057, | |
| "step": 3095, | |
| "valid_targets_mean": 5209.5, | |
| "valid_targets_min": 1934 | |
| }, | |
| { | |
| "epoch": 4.952076677316294, | |
| "grad_norm": 0.42040290394596574, | |
| "learning_rate": 9.570756365311418e-06, | |
| "loss": 0.1412, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1560743749141693, | |
| "step": 3100, | |
| "valid_targets_mean": 4999.1, | |
| "valid_targets_min": 1129 | |
| }, | |
| { | |
| "epoch": 4.960063897763578, | |
| "grad_norm": 0.4228988525636665, | |
| "learning_rate": 9.502854505210326e-06, | |
| "loss": 0.132, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12458611279726028, | |
| "step": 3105, | |
| "valid_targets_mean": 4566.2, | |
| "valid_targets_min": 966 | |
| }, | |
| { | |
| "epoch": 4.968051118210862, | |
| "grad_norm": 0.49618214601387967, | |
| "learning_rate": 9.435119238145018e-06, | |
| "loss": 0.1401, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13144153356552124, | |
| "step": 3110, | |
| "valid_targets_mean": 4248.2, | |
| "valid_targets_min": 736 | |
| }, | |
| { | |
| "epoch": 4.976038338658147, | |
| "grad_norm": 0.4297986513976899, | |
| "learning_rate": 9.367551639095704e-06, | |
| "loss": 0.1448, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14683717489242554, | |
| "step": 3115, | |
| "valid_targets_mean": 4752.8, | |
| "valid_targets_min": 1627 | |
| }, | |
| { | |
| "epoch": 4.984025559105431, | |
| "grad_norm": 0.476708428928287, | |
| "learning_rate": 9.300152780381664e-06, | |
| "loss": 0.1507, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15097108483314514, | |
| "step": 3120, | |
| "valid_targets_mean": 5958.0, | |
| "valid_targets_min": 3142 | |
| }, | |
| { | |
| "epoch": 4.992012779552716, | |
| "grad_norm": 0.37717889001385313, | |
| "learning_rate": 9.232923731644216e-06, | |
| "loss": 0.1338, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1160600483417511, | |
| "step": 3125, | |
| "valid_targets_mean": 5766.6, | |
| "valid_targets_min": 3233 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 0.4145923303573134, | |
| "learning_rate": 9.16586555982972e-06, | |
| "loss": 0.1422, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13584575057029724, | |
| "step": 3130, | |
| "valid_targets_mean": 5288.1, | |
| "valid_targets_min": 1868 | |
| }, | |
| { | |
| "epoch": 5.007987220447284, | |
| "grad_norm": 0.3996496780407897, | |
| "learning_rate": 9.098979329172702e-06, | |
| "loss": 0.1362, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14415614306926727, | |
| "step": 3135, | |
| "valid_targets_mean": 5790.2, | |
| "valid_targets_min": 1878 | |
| }, | |
| { | |
| "epoch": 5.015974440894569, | |
| "grad_norm": 0.36607583997858123, | |
| "learning_rate": 9.032266101178872e-06, | |
| "loss": 0.1224, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10579011589288712, | |
| "step": 3140, | |
| "valid_targets_mean": 5548.6, | |
| "valid_targets_min": 2620 | |
| }, | |
| { | |
| "epoch": 5.023961661341853, | |
| "grad_norm": 0.42407706221719077, | |
| "learning_rate": 8.965726934608392e-06, | |
| "loss": 0.1261, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11896810680627823, | |
| "step": 3145, | |
| "valid_targets_mean": 5135.3, | |
| "valid_targets_min": 2005 | |
| }, | |
| { | |
| "epoch": 5.031948881789138, | |
| "grad_norm": 0.391814299507267, | |
| "learning_rate": 8.899362885458964e-06, | |
| "loss": 0.1352, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11073155701160431, | |
| "step": 3150, | |
| "valid_targets_mean": 5157.8, | |
| "valid_targets_min": 2616 | |
| }, | |
| { | |
| "epoch": 5.039936102236422, | |
| "grad_norm": 0.4389079217597968, | |
| "learning_rate": 8.833175006949165e-06, | |
| "loss": 0.1261, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11496037989854813, | |
| "step": 3155, | |
| "valid_targets_mean": 4873.9, | |
| "valid_targets_min": 822 | |
| }, | |
| { | |
| "epoch": 5.047923322683706, | |
| "grad_norm": 0.45420730578563395, | |
| "learning_rate": 8.767164349501648e-06, | |
| "loss": 0.1306, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13046647608280182, | |
| "step": 3160, | |
| "valid_targets_mean": 5349.7, | |
| "valid_targets_min": 1667 | |
| }, | |
| { | |
| "epoch": 5.055910543130991, | |
| "grad_norm": 0.4270028436404705, | |
| "learning_rate": 8.70133196072654e-06, | |
| "loss": 0.1433, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13287319242954254, | |
| "step": 3165, | |
| "valid_targets_mean": 4957.7, | |
| "valid_targets_min": 1093 | |
| }, | |
| { | |
| "epoch": 5.063897763578275, | |
| "grad_norm": 0.44769757237954166, | |
| "learning_rate": 8.63567888540478e-06, | |
| "loss": 0.1357, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16756519675254822, | |
| "step": 3170, | |
| "valid_targets_mean": 5475.4, | |
| "valid_targets_min": 794 | |
| }, | |
| { | |
| "epoch": 5.0718849840255595, | |
| "grad_norm": 0.43416183051660023, | |
| "learning_rate": 8.570206165471535e-06, | |
| "loss": 0.1221, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10582110285758972, | |
| "step": 3175, | |
| "valid_targets_mean": 5131.0, | |
| "valid_targets_min": 2792 | |
| }, | |
| { | |
| "epoch": 5.079872204472843, | |
| "grad_norm": 0.49261163257102736, | |
| "learning_rate": 8.504914839999691e-06, | |
| "loss": 0.1339, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14795424044132233, | |
| "step": 3180, | |
| "valid_targets_mean": 5218.9, | |
| "valid_targets_min": 2270 | |
| }, | |
| { | |
| "epoch": 5.087859424920127, | |
| "grad_norm": 0.3941568181771655, | |
| "learning_rate": 8.439805945183333e-06, | |
| "loss": 0.1236, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12936750054359436, | |
| "step": 3185, | |
| "valid_targets_mean": 5971.9, | |
| "valid_targets_min": 2729 | |
| }, | |
| { | |
| "epoch": 5.095846645367412, | |
| "grad_norm": 0.4243186553880082, | |
| "learning_rate": 8.374880514321318e-06, | |
| "loss": 0.1326, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14162969589233398, | |
| "step": 3190, | |
| "valid_targets_mean": 5262.4, | |
| "valid_targets_min": 1129 | |
| }, | |
| { | |
| "epoch": 5.103833865814696, | |
| "grad_norm": 0.4326438938142632, | |
| "learning_rate": 8.310139577800864e-06, | |
| "loss": 0.1345, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10833601653575897, | |
| "step": 3195, | |
| "valid_targets_mean": 4824.5, | |
| "valid_targets_min": 2319 | |
| }, | |
| { | |
| "epoch": 5.111821086261981, | |
| "grad_norm": 0.43753208972217505, | |
| "learning_rate": 8.245584163081228e-06, | |
| "loss": 0.125, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11919113993644714, | |
| "step": 3200, | |
| "valid_targets_mean": 4601.9, | |
| "valid_targets_min": 2552 | |
| }, | |
| { | |
| "epoch": 5.119808306709265, | |
| "grad_norm": 0.4425755216974976, | |
| "learning_rate": 8.18121529467735e-06, | |
| "loss": 0.13, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11845006048679352, | |
| "step": 3205, | |
| "valid_targets_mean": 5014.2, | |
| "valid_targets_min": 2470 | |
| }, | |
| { | |
| "epoch": 5.127795527156549, | |
| "grad_norm": 0.4257623559583295, | |
| "learning_rate": 8.117033994143635e-06, | |
| "loss": 0.1279, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13290351629257202, | |
| "step": 3210, | |
| "valid_targets_mean": 5347.4, | |
| "valid_targets_min": 2245 | |
| }, | |
| { | |
| "epoch": 5.135782747603834, | |
| "grad_norm": 0.4308505974079372, | |
| "learning_rate": 8.053041280057733e-06, | |
| "loss": 0.1228, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12126690149307251, | |
| "step": 3215, | |
| "valid_targets_mean": 4555.8, | |
| "valid_targets_min": 1008 | |
| }, | |
| { | |
| "epoch": 5.143769968051118, | |
| "grad_norm": 0.4189383686117709, | |
| "learning_rate": 7.989238168004347e-06, | |
| "loss": 0.1238, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12490759789943695, | |
| "step": 3220, | |
| "valid_targets_mean": 6104.8, | |
| "valid_targets_min": 3254 | |
| }, | |
| { | |
| "epoch": 5.151757188498403, | |
| "grad_norm": 0.4332250495317117, | |
| "learning_rate": 7.925625670559158e-06, | |
| "loss": 0.1352, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12067738175392151, | |
| "step": 3225, | |
| "valid_targets_mean": 5199.1, | |
| "valid_targets_min": 1610 | |
| }, | |
| { | |
| "epoch": 5.159744408945687, | |
| "grad_norm": 0.42692430751048394, | |
| "learning_rate": 7.862204797272716e-06, | |
| "loss": 0.139, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14403921365737915, | |
| "step": 3230, | |
| "valid_targets_mean": 5759.9, | |
| "valid_targets_min": 2536 | |
| }, | |
| { | |
| "epoch": 5.167731629392971, | |
| "grad_norm": 0.4329505281595029, | |
| "learning_rate": 7.798976554654438e-06, | |
| "loss": 0.1305, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11265379190444946, | |
| "step": 3235, | |
| "valid_targets_mean": 4747.3, | |
| "valid_targets_min": 1277 | |
| }, | |
| { | |
| "epoch": 5.175718849840256, | |
| "grad_norm": 0.4463908820514333, | |
| "learning_rate": 7.73594194615663e-06, | |
| "loss": 0.1331, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1566663384437561, | |
| "step": 3240, | |
| "valid_targets_mean": 5206.6, | |
| "valid_targets_min": 1309 | |
| }, | |
| { | |
| "epoch": 5.18370607028754, | |
| "grad_norm": 0.4398349209256662, | |
| "learning_rate": 7.673101972158572e-06, | |
| "loss": 0.135, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11558836698532104, | |
| "step": 3245, | |
| "valid_targets_mean": 5067.0, | |
| "valid_targets_min": 2267 | |
| }, | |
| { | |
| "epoch": 5.1916932907348246, | |
| "grad_norm": 0.4099745621174123, | |
| "learning_rate": 7.610457629950621e-06, | |
| "loss": 0.1311, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1297166347503662, | |
| "step": 3250, | |
| "valid_targets_mean": 5251.2, | |
| "valid_targets_min": 3080 | |
| }, | |
| { | |
| "epoch": 5.199680511182109, | |
| "grad_norm": 0.4875957508996389, | |
| "learning_rate": 7.548009913718402e-06, | |
| "loss": 0.1269, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.134623721241951, | |
| "step": 3255, | |
| "valid_targets_mean": 5723.5, | |
| "valid_targets_min": 2671 | |
| }, | |
| { | |
| "epoch": 5.207667731629393, | |
| "grad_norm": 0.3903879002897779, | |
| "learning_rate": 7.485759814527034e-06, | |
| "loss": 0.1304, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11780675500631332, | |
| "step": 3260, | |
| "valid_targets_mean": 5756.2, | |
| "valid_targets_min": 2539 | |
| }, | |
| { | |
| "epoch": 5.215654952076678, | |
| "grad_norm": 0.432265041485957, | |
| "learning_rate": 7.423708320305361e-06, | |
| "loss": 0.1297, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12412317842245102, | |
| "step": 3265, | |
| "valid_targets_mean": 4916.5, | |
| "valid_targets_min": 2164 | |
| }, | |
| { | |
| "epoch": 5.223642172523961, | |
| "grad_norm": 0.4796132619202076, | |
| "learning_rate": 7.361856415830335e-06, | |
| "loss": 0.136, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14176784455776215, | |
| "step": 3270, | |
| "valid_targets_mean": 4793.1, | |
| "valid_targets_min": 1520 | |
| }, | |
| { | |
| "epoch": 5.231629392971246, | |
| "grad_norm": 0.4147163044728787, | |
| "learning_rate": 7.300205082711336e-06, | |
| "loss": 0.1233, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10913220047950745, | |
| "step": 3275, | |
| "valid_targets_mean": 5197.5, | |
| "valid_targets_min": 2858 | |
| }, | |
| { | |
| "epoch": 5.23961661341853, | |
| "grad_norm": 0.43707733192346127, | |
| "learning_rate": 7.2387552993746204e-06, | |
| "loss": 0.1256, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10962367057800293, | |
| "step": 3280, | |
| "valid_targets_mean": 4511.2, | |
| "valid_targets_min": 2345 | |
| }, | |
| { | |
| "epoch": 5.247603833865814, | |
| "grad_norm": 0.46403682669659124, | |
| "learning_rate": 7.177508041047769e-06, | |
| "loss": 0.1326, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12127917259931564, | |
| "step": 3285, | |
| "valid_targets_mean": 5016.0, | |
| "valid_targets_min": 1505 | |
| }, | |
| { | |
| "epoch": 5.255591054313099, | |
| "grad_norm": 0.41905186853610676, | |
| "learning_rate": 7.116464279744262e-06, | |
| "loss": 0.1217, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1214476227760315, | |
| "step": 3290, | |
| "valid_targets_mean": 5025.4, | |
| "valid_targets_min": 3347 | |
| }, | |
| { | |
| "epoch": 5.263578274760383, | |
| "grad_norm": 0.4984331131561831, | |
| "learning_rate": 7.055624984247977e-06, | |
| "loss": 0.1357, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.131068155169487, | |
| "step": 3295, | |
| "valid_targets_mean": 4773.6, | |
| "valid_targets_min": 1805 | |
| }, | |
| { | |
| "epoch": 5.271565495207668, | |
| "grad_norm": 0.42633292065820244, | |
| "learning_rate": 6.9949911200978805e-06, | |
| "loss": 0.1331, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14695999026298523, | |
| "step": 3300, | |
| "valid_targets_mean": 5357.7, | |
| "valid_targets_min": 2337 | |
| }, | |
| { | |
| "epoch": 5.279552715654952, | |
| "grad_norm": 0.4385631436029575, | |
| "learning_rate": 6.934563649572678e-06, | |
| "loss": 0.1313, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11017829179763794, | |
| "step": 3305, | |
| "valid_targets_mean": 4488.4, | |
| "valid_targets_min": 1783 | |
| }, | |
| { | |
| "epoch": 5.287539936102236, | |
| "grad_norm": 0.463161443429704, | |
| "learning_rate": 6.874343531675521e-06, | |
| "loss": 0.1313, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.150054931640625, | |
| "step": 3310, | |
| "valid_targets_mean": 4672.8, | |
| "valid_targets_min": 597 | |
| }, | |
| { | |
| "epoch": 5.295527156549521, | |
| "grad_norm": 0.43005990166774005, | |
| "learning_rate": 6.814331722118837e-06, | |
| "loss": 0.1338, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14076587557792664, | |
| "step": 3315, | |
| "valid_targets_mean": 5725.1, | |
| "valid_targets_min": 1859 | |
| }, | |
| { | |
| "epoch": 5.303514376996805, | |
| "grad_norm": 0.45237370010427796, | |
| "learning_rate": 6.754529173309112e-06, | |
| "loss": 0.1301, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11547715961933136, | |
| "step": 3320, | |
| "valid_targets_mean": 5078.8, | |
| "valid_targets_min": 2912 | |
| }, | |
| { | |
| "epoch": 5.31150159744409, | |
| "grad_norm": 0.44502199228614653, | |
| "learning_rate": 6.694936834331809e-06, | |
| "loss": 0.1196, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12647053599357605, | |
| "step": 3325, | |
| "valid_targets_mean": 5275.9, | |
| "valid_targets_min": 2599 | |
| }, | |
| { | |
| "epoch": 5.319488817891374, | |
| "grad_norm": 0.5602905216304396, | |
| "learning_rate": 6.635555650936278e-06, | |
| "loss": 0.1375, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11837688088417053, | |
| "step": 3330, | |
| "valid_targets_mean": 4813.0, | |
| "valid_targets_min": 1514 | |
| }, | |
| { | |
| "epoch": 5.327476038338658, | |
| "grad_norm": 0.45505112639439443, | |
| "learning_rate": 6.576386565520794e-06, | |
| "loss": 0.1401, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15331462025642395, | |
| "step": 3335, | |
| "valid_targets_mean": 5086.4, | |
| "valid_targets_min": 3067 | |
| }, | |
| { | |
| "epoch": 5.335463258785943, | |
| "grad_norm": 0.4544496565614777, | |
| "learning_rate": 6.5174305171175336e-06, | |
| "loss": 0.1302, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13496166467666626, | |
| "step": 3340, | |
| "valid_targets_mean": 4816.9, | |
| "valid_targets_min": 1222 | |
| }, | |
| { | |
| "epoch": 5.343450479233227, | |
| "grad_norm": 0.46403407365621835, | |
| "learning_rate": 6.458688441377734e-06, | |
| "loss": 0.1322, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14024507999420166, | |
| "step": 3345, | |
| "valid_targets_mean": 4533.0, | |
| "valid_targets_min": 399 | |
| }, | |
| { | |
| "epoch": 5.3514376996805115, | |
| "grad_norm": 0.4521234617595679, | |
| "learning_rate": 6.400161270556817e-06, | |
| "loss": 0.1336, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1282578706741333, | |
| "step": 3350, | |
| "valid_targets_mean": 4508.7, | |
| "valid_targets_min": 275 | |
| }, | |
| { | |
| "epoch": 5.359424920127796, | |
| "grad_norm": 0.44261670269535724, | |
| "learning_rate": 6.341849933499573e-06, | |
| "loss": 0.1244, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12378348410129547, | |
| "step": 3355, | |
| "valid_targets_mean": 4897.5, | |
| "valid_targets_min": 1438 | |
| }, | |
| { | |
| "epoch": 5.36741214057508, | |
| "grad_norm": 0.42552393299511426, | |
| "learning_rate": 6.283755355625472e-06, | |
| "loss": 0.1377, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11719924211502075, | |
| "step": 3360, | |
| "valid_targets_mean": 5004.2, | |
| "valid_targets_min": 2266 | |
| }, | |
| { | |
| "epoch": 5.375399361022364, | |
| "grad_norm": 0.44170728187372643, | |
| "learning_rate": 6.225878458913917e-06, | |
| "loss": 0.1307, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13522031903266907, | |
| "step": 3365, | |
| "valid_targets_mean": 5598.8, | |
| "valid_targets_min": 3197 | |
| }, | |
| { | |
| "epoch": 5.383386581469648, | |
| "grad_norm": 0.5061548909266773, | |
| "learning_rate": 6.168220161889678e-06, | |
| "loss": 0.1254, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1257493644952774, | |
| "step": 3370, | |
| "valid_targets_mean": 4849.5, | |
| "valid_targets_min": 2170 | |
| }, | |
| { | |
| "epoch": 5.391373801916933, | |
| "grad_norm": 0.41610193838589155, | |
| "learning_rate": 6.110781379608226e-06, | |
| "loss": 0.1255, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10494233667850494, | |
| "step": 3375, | |
| "valid_targets_mean": 5387.9, | |
| "valid_targets_min": 1916 | |
| }, | |
| { | |
| "epoch": 5.399361022364217, | |
| "grad_norm": 0.4076940777306394, | |
| "learning_rate": 6.053563023641318e-06, | |
| "loss": 0.1348, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12908264994621277, | |
| "step": 3380, | |
| "valid_targets_mean": 5463.6, | |
| "valid_targets_min": 2897 | |
| }, | |
| { | |
| "epoch": 5.407348242811501, | |
| "grad_norm": 0.43707435415564755, | |
| "learning_rate": 5.996566002062439e-06, | |
| "loss": 0.1299, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13832756876945496, | |
| "step": 3385, | |
| "valid_targets_mean": 5148.5, | |
| "valid_targets_min": 2363 | |
| }, | |
| { | |
| "epoch": 5.415335463258786, | |
| "grad_norm": 0.424775163843778, | |
| "learning_rate": 5.939791219432442e-06, | |
| "loss": 0.1331, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11907709389925003, | |
| "step": 3390, | |
| "valid_targets_mean": 5787.8, | |
| "valid_targets_min": 3185 | |
| }, | |
| { | |
| "epoch": 5.42332268370607, | |
| "grad_norm": 0.471019307609933, | |
| "learning_rate": 5.8832395767851846e-06, | |
| "loss": 0.1263, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1155877411365509, | |
| "step": 3395, | |
| "valid_targets_mean": 4662.1, | |
| "valid_targets_min": 1368 | |
| }, | |
| { | |
| "epoch": 5.431309904153355, | |
| "grad_norm": 0.4550644292377657, | |
| "learning_rate": 5.826911971613198e-06, | |
| "loss": 0.1379, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14119279384613037, | |
| "step": 3400, | |
| "valid_targets_mean": 5145.4, | |
| "valid_targets_min": 2618 | |
| }, | |
| { | |
| "epoch": 5.439297124600639, | |
| "grad_norm": 0.41524027617107767, | |
| "learning_rate": 5.770809297853495e-06, | |
| "loss": 0.1291, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1215624213218689, | |
| "step": 3405, | |
| "valid_targets_mean": 5328.4, | |
| "valid_targets_min": 2245 | |
| }, | |
| { | |
| "epoch": 5.447284345047923, | |
| "grad_norm": 0.45085680558786, | |
| "learning_rate": 5.714932445873325e-06, | |
| "loss": 0.127, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11698678135871887, | |
| "step": 3410, | |
| "valid_targets_mean": 5146.0, | |
| "valid_targets_min": 1780 | |
| }, | |
| { | |
| "epoch": 5.455271565495208, | |
| "grad_norm": 0.36306472132135814, | |
| "learning_rate": 5.659282302456115e-06, | |
| "loss": 0.1258, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11686362326145172, | |
| "step": 3415, | |
| "valid_targets_mean": 6465.1, | |
| "valid_targets_min": 2635 | |
| }, | |
| { | |
| "epoch": 5.463258785942492, | |
| "grad_norm": 0.4861913930837589, | |
| "learning_rate": 5.603859750787302e-06, | |
| "loss": 0.1331, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15752384066581726, | |
| "step": 3420, | |
| "valid_targets_mean": 5122.1, | |
| "valid_targets_min": 2449 | |
| }, | |
| { | |
| "epoch": 5.4712460063897765, | |
| "grad_norm": 0.5116564623917675, | |
| "learning_rate": 5.548665670440418e-06, | |
| "loss": 0.1345, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13592877984046936, | |
| "step": 3425, | |
| "valid_targets_mean": 4740.7, | |
| "valid_targets_min": 2428 | |
| }, | |
| { | |
| "epoch": 5.479233226837061, | |
| "grad_norm": 0.4420467245666536, | |
| "learning_rate": 5.4937009373630535e-06, | |
| "loss": 0.1292, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14905400574207306, | |
| "step": 3430, | |
| "valid_targets_mean": 5819.4, | |
| "valid_targets_min": 2509 | |
| }, | |
| { | |
| "epoch": 5.487220447284345, | |
| "grad_norm": 0.42001287509088997, | |
| "learning_rate": 5.438966423862997e-06, | |
| "loss": 0.1334, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12227460741996765, | |
| "step": 3435, | |
| "valid_targets_mean": 5873.6, | |
| "valid_targets_min": 2993 | |
| }, | |
| { | |
| "epoch": 5.49520766773163, | |
| "grad_norm": 0.5002417350096667, | |
| "learning_rate": 5.384462998594384e-06, | |
| "loss": 0.1305, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14182063937187195, | |
| "step": 3440, | |
| "valid_targets_mean": 4336.8, | |
| "valid_targets_min": 1158 | |
| }, | |
| { | |
| "epoch": 5.503194888178914, | |
| "grad_norm": 0.44178449766828326, | |
| "learning_rate": 5.330191526543884e-06, | |
| "loss": 0.1257, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1353127509355545, | |
| "step": 3445, | |
| "valid_targets_mean": 5306.8, | |
| "valid_targets_min": 1974 | |
| }, | |
| { | |
| "epoch": 5.511182108626198, | |
| "grad_norm": 0.4473454269016277, | |
| "learning_rate": 5.276152869017028e-06, | |
| "loss": 0.1462, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16462580859661102, | |
| "step": 3450, | |
| "valid_targets_mean": 5123.2, | |
| "valid_targets_min": 3357 | |
| }, | |
| { | |
| "epoch": 5.519169329073483, | |
| "grad_norm": 0.4142802816566718, | |
| "learning_rate": 5.22234788362447e-06, | |
| "loss": 0.1311, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1246831938624382, | |
| "step": 3455, | |
| "valid_targets_mean": 5452.1, | |
| "valid_targets_min": 1707 | |
| }, | |
| { | |
| "epoch": 5.527156549520766, | |
| "grad_norm": 0.4286401429495573, | |
| "learning_rate": 5.168777424268454e-06, | |
| "loss": 0.1356, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13388681411743164, | |
| "step": 3460, | |
| "valid_targets_mean": 5635.9, | |
| "valid_targets_min": 2705 | |
| }, | |
| { | |
| "epoch": 5.535143769968051, | |
| "grad_norm": 0.4328975064991749, | |
| "learning_rate": 5.115442341129171e-06, | |
| "loss": 0.1349, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11813346296548843, | |
| "step": 3465, | |
| "valid_targets_mean": 5013.9, | |
| "valid_targets_min": 641 | |
| }, | |
| { | |
| "epoch": 5.543130990415335, | |
| "grad_norm": 0.5180471284346164, | |
| "learning_rate": 5.062343480651364e-06, | |
| "loss": 0.1316, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11512558907270432, | |
| "step": 3470, | |
| "valid_targets_mean": 4785.0, | |
| "valid_targets_min": 416 | |
| }, | |
| { | |
| "epoch": 5.55111821086262, | |
| "grad_norm": 0.41608403073691597, | |
| "learning_rate": 5.009481685530817e-06, | |
| "loss": 0.1263, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1358884572982788, | |
| "step": 3475, | |
| "valid_targets_mean": 5345.8, | |
| "valid_targets_min": 315 | |
| }, | |
| { | |
| "epoch": 5.559105431309904, | |
| "grad_norm": 0.40862554554517955, | |
| "learning_rate": 4.956857794701026e-06, | |
| "loss": 0.1305, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11328987777233124, | |
| "step": 3480, | |
| "valid_targets_mean": 5166.4, | |
| "valid_targets_min": 627 | |
| }, | |
| { | |
| "epoch": 5.567092651757188, | |
| "grad_norm": 0.4488789839306291, | |
| "learning_rate": 4.904472643319873e-06, | |
| "loss": 0.1313, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14518974721431732, | |
| "step": 3485, | |
| "valid_targets_mean": 5571.2, | |
| "valid_targets_min": 1374 | |
| }, | |
| { | |
| "epoch": 5.575079872204473, | |
| "grad_norm": 0.46397180679702854, | |
| "learning_rate": 4.852327062756352e-06, | |
| "loss": 0.1367, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1412065327167511, | |
| "step": 3490, | |
| "valid_targets_mean": 5190.4, | |
| "valid_targets_min": 1571 | |
| }, | |
| { | |
| "epoch": 5.583067092651757, | |
| "grad_norm": 0.4340299374186009, | |
| "learning_rate": 4.800421880577411e-06, | |
| "loss": 0.1326, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11852262914180756, | |
| "step": 3495, | |
| "valid_targets_mean": 5491.2, | |
| "valid_targets_min": 2944 | |
| }, | |
| { | |
| "epoch": 5.5910543130990416, | |
| "grad_norm": 0.4395392158236446, | |
| "learning_rate": 4.748757920534779e-06, | |
| "loss": 0.1401, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14287468791007996, | |
| "step": 3500, | |
| "valid_targets_mean": 5008.6, | |
| "valid_targets_min": 2339 | |
| }, | |
| { | |
| "epoch": 5.599041533546326, | |
| "grad_norm": 0.4617324502694257, | |
| "learning_rate": 4.697336002551947e-06, | |
| "loss": 0.1321, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11883892118930817, | |
| "step": 3505, | |
| "valid_targets_mean": 5560.0, | |
| "valid_targets_min": 374 | |
| }, | |
| { | |
| "epoch": 5.60702875399361, | |
| "grad_norm": 0.4892467829853935, | |
| "learning_rate": 4.6461569427110684e-06, | |
| "loss": 0.1299, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10661804676055908, | |
| "step": 3510, | |
| "valid_targets_mean": 4936.2, | |
| "valid_targets_min": 1133 | |
| }, | |
| { | |
| "epoch": 5.615015974440895, | |
| "grad_norm": 0.46519124203936624, | |
| "learning_rate": 4.5952215532401146e-06, | |
| "loss": 0.1365, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15909770131111145, | |
| "step": 3515, | |
| "valid_targets_mean": 4965.2, | |
| "valid_targets_min": 236 | |
| }, | |
| { | |
| "epoch": 5.623003194888179, | |
| "grad_norm": 0.4144087349402236, | |
| "learning_rate": 4.544530642499894e-06, | |
| "loss": 0.1416, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1373518705368042, | |
| "step": 3520, | |
| "valid_targets_mean": 5098.4, | |
| "valid_targets_min": 1452 | |
| }, | |
| { | |
| "epoch": 5.6309904153354635, | |
| "grad_norm": 0.4535396178895797, | |
| "learning_rate": 4.4940850149712765e-06, | |
| "loss": 0.1201, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12060095369815826, | |
| "step": 3525, | |
| "valid_targets_mean": 4842.2, | |
| "valid_targets_min": 1691 | |
| }, | |
| { | |
| "epoch": 5.638977635782748, | |
| "grad_norm": 0.5020002802088452, | |
| "learning_rate": 4.443885471242418e-06, | |
| "loss": 0.1313, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15742410719394684, | |
| "step": 3530, | |
| "valid_targets_mean": 4325.4, | |
| "valid_targets_min": 918 | |
| }, | |
| { | |
| "epoch": 5.646964856230032, | |
| "grad_norm": 0.44209419159312086, | |
| "learning_rate": 4.393932807996017e-06, | |
| "loss": 0.1219, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1325576901435852, | |
| "step": 3535, | |
| "valid_targets_mean": 5183.6, | |
| "valid_targets_min": 3320 | |
| }, | |
| { | |
| "epoch": 5.654952076677317, | |
| "grad_norm": 0.43378501377546835, | |
| "learning_rate": 4.344227817996735e-06, | |
| "loss": 0.1389, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12083230912685394, | |
| "step": 3540, | |
| "valid_targets_mean": 5181.2, | |
| "valid_targets_min": 783 | |
| }, | |
| { | |
| "epoch": 5.6629392971246, | |
| "grad_norm": 0.5161310130883322, | |
| "learning_rate": 4.294771290078548e-06, | |
| "loss": 0.129, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13089406490325928, | |
| "step": 3545, | |
| "valid_targets_mean": 5580.3, | |
| "valid_targets_min": 2808 | |
| }, | |
| { | |
| "epoch": 5.6709265175718855, | |
| "grad_norm": 0.576118536628751, | |
| "learning_rate": 4.245564009132293e-06, | |
| "loss": 0.1366, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1379978209733963, | |
| "step": 3550, | |
| "valid_targets_mean": 5044.8, | |
| "valid_targets_min": 1512 | |
| }, | |
| { | |
| "epoch": 5.678913738019169, | |
| "grad_norm": 0.39914383089678107, | |
| "learning_rate": 4.196606756093138e-06, | |
| "loss": 0.1283, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11285041272640228, | |
| "step": 3555, | |
| "valid_targets_mean": 6019.3, | |
| "valid_targets_min": 3357 | |
| }, | |
| { | |
| "epoch": 5.686900958466453, | |
| "grad_norm": 0.46134426630128794, | |
| "learning_rate": 4.147900307928268e-06, | |
| "loss": 0.1354, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14031925797462463, | |
| "step": 3560, | |
| "valid_targets_mean": 4953.0, | |
| "valid_targets_min": 1941 | |
| }, | |
| { | |
| "epoch": 5.694888178913738, | |
| "grad_norm": 0.44519797510119424, | |
| "learning_rate": 4.099445437624487e-06, | |
| "loss": 0.1269, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14868547022342682, | |
| "step": 3565, | |
| "valid_targets_mean": 5354.8, | |
| "valid_targets_min": 2360 | |
| }, | |
| { | |
| "epoch": 5.702875399361022, | |
| "grad_norm": 0.4857647413284097, | |
| "learning_rate": 4.051242914175995e-06, | |
| "loss": 0.1283, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1460459977388382, | |
| "step": 3570, | |
| "valid_targets_mean": 4833.9, | |
| "valid_targets_min": 1866 | |
| }, | |
| { | |
| "epoch": 5.710862619808307, | |
| "grad_norm": 0.4045252768491045, | |
| "learning_rate": 4.003293502572163e-06, | |
| "loss": 0.1306, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11539411544799805, | |
| "step": 3575, | |
| "valid_targets_mean": 5423.4, | |
| "valid_targets_min": 1744 | |
| }, | |
| { | |
| "epoch": 5.718849840255591, | |
| "grad_norm": 0.4280757439237818, | |
| "learning_rate": 3.955597963785391e-06, | |
| "loss": 0.1254, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12244729697704315, | |
| "step": 3580, | |
| "valid_targets_mean": 5413.9, | |
| "valid_targets_min": 2924 | |
| }, | |
| { | |
| "epoch": 5.726837060702875, | |
| "grad_norm": 0.4261277071252784, | |
| "learning_rate": 3.908157054759048e-06, | |
| "loss": 0.1412, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15924279391765594, | |
| "step": 3585, | |
| "valid_targets_mean": 5404.3, | |
| "valid_targets_min": 3324 | |
| }, | |
| { | |
| "epoch": 5.73482428115016, | |
| "grad_norm": 0.4801327791783322, | |
| "learning_rate": 3.860971528395427e-06, | |
| "loss": 0.1261, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13970091938972473, | |
| "step": 3590, | |
| "valid_targets_mean": 4623.3, | |
| "valid_targets_min": 779 | |
| }, | |
| { | |
| "epoch": 5.742811501597444, | |
| "grad_norm": 0.4468320609852807, | |
| "learning_rate": 3.814042133543847e-06, | |
| "loss": 0.1448, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14838215708732605, | |
| "step": 3595, | |
| "valid_targets_mean": 6140.8, | |
| "valid_targets_min": 2443 | |
| }, | |
| { | |
| "epoch": 5.7507987220447285, | |
| "grad_norm": 0.5766098241929758, | |
| "learning_rate": 3.7673696149887117e-06, | |
| "loss": 0.1249, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1346101462841034, | |
| "step": 3600, | |
| "valid_targets_mean": 5091.9, | |
| "valid_targets_min": 1571 | |
| }, | |
| { | |
| "epoch": 5.758785942492013, | |
| "grad_norm": 0.49034943466324993, | |
| "learning_rate": 3.72095471343773e-06, | |
| "loss": 0.1447, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15312346816062927, | |
| "step": 3605, | |
| "valid_targets_mean": 4705.5, | |
| "valid_targets_min": 1276 | |
| }, | |
| { | |
| "epoch": 5.766773162939297, | |
| "grad_norm": 0.4841334583191147, | |
| "learning_rate": 3.674798165510136e-06, | |
| "loss": 0.133, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1331675797700882, | |
| "step": 3610, | |
| "valid_targets_mean": 4928.2, | |
| "valid_targets_min": 1600 | |
| }, | |
| { | |
| "epoch": 5.774760383386582, | |
| "grad_norm": 0.41277416680792434, | |
| "learning_rate": 3.6289007037250244e-06, | |
| "loss": 0.1205, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10908909887075424, | |
| "step": 3615, | |
| "valid_targets_mean": 4816.3, | |
| "valid_targets_min": 1619 | |
| }, | |
| { | |
| "epoch": 5.782747603833866, | |
| "grad_norm": 0.4761587674247828, | |
| "learning_rate": 3.5832630564897073e-06, | |
| "loss": 0.1301, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1364089995622635, | |
| "step": 3620, | |
| "valid_targets_mean": 4879.2, | |
| "valid_targets_min": 2515 | |
| }, | |
| { | |
| "epoch": 5.7907348242811505, | |
| "grad_norm": 0.4428949210282417, | |
| "learning_rate": 3.5378859480881443e-06, | |
| "loss": 0.1349, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13722603023052216, | |
| "step": 3625, | |
| "valid_targets_mean": 5562.2, | |
| "valid_targets_min": 2746 | |
| }, | |
| { | |
| "epoch": 5.798722044728435, | |
| "grad_norm": 0.4001563456292007, | |
| "learning_rate": 3.492770098669478e-06, | |
| "loss": 0.1399, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10774213820695877, | |
| "step": 3630, | |
| "valid_targets_mean": 5270.2, | |
| "valid_targets_min": 2006 | |
| }, | |
| { | |
| "epoch": 5.806709265175719, | |
| "grad_norm": 0.4436488334607921, | |
| "learning_rate": 3.4479162242365717e-06, | |
| "loss": 0.1276, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11889493465423584, | |
| "step": 3635, | |
| "valid_targets_mean": 5386.1, | |
| "valid_targets_min": 966 | |
| }, | |
| { | |
| "epoch": 5.814696485623003, | |
| "grad_norm": 0.4626765054212039, | |
| "learning_rate": 3.403325036634679e-06, | |
| "loss": 0.1343, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12846167385578156, | |
| "step": 3640, | |
| "valid_targets_mean": 4516.9, | |
| "valid_targets_min": 2323 | |
| }, | |
| { | |
| "epoch": 5.822683706070287, | |
| "grad_norm": 0.44135786249250614, | |
| "learning_rate": 3.3589972435401184e-06, | |
| "loss": 0.1408, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13893258571624756, | |
| "step": 3645, | |
| "valid_targets_mean": 4899.9, | |
| "valid_targets_min": 1167 | |
| }, | |
| { | |
| "epoch": 5.830670926517572, | |
| "grad_norm": 0.4004779146241053, | |
| "learning_rate": 3.3149335484490553e-06, | |
| "loss": 0.1284, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12162289023399353, | |
| "step": 3650, | |
| "valid_targets_mean": 5508.4, | |
| "valid_targets_min": 2368 | |
| }, | |
| { | |
| "epoch": 5.838658146964856, | |
| "grad_norm": 0.4398288330197614, | |
| "learning_rate": 3.2711346506663346e-06, | |
| "loss": 0.1335, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12186942994594574, | |
| "step": 3655, | |
| "valid_targets_mean": 4510.9, | |
| "valid_targets_min": 2253 | |
| }, | |
| { | |
| "epoch": 5.84664536741214, | |
| "grad_norm": 0.43086529481980373, | |
| "learning_rate": 3.2276012452943893e-06, | |
| "loss": 0.1209, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11107099801301956, | |
| "step": 3660, | |
| "valid_targets_mean": 5440.8, | |
| "valid_targets_min": 2601 | |
| }, | |
| { | |
| "epoch": 5.854632587859425, | |
| "grad_norm": 0.41815726745877874, | |
| "learning_rate": 3.1843340232222064e-06, | |
| "loss": 0.1342, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12301813811063766, | |
| "step": 3665, | |
| "valid_targets_mean": 6054.9, | |
| "valid_targets_min": 2170 | |
| }, | |
| { | |
| "epoch": 5.862619808306709, | |
| "grad_norm": 0.4841406329823737, | |
| "learning_rate": 3.1413336711143437e-06, | |
| "loss": 0.1297, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1188805103302002, | |
| "step": 3670, | |
| "valid_targets_mean": 4153.4, | |
| "valid_targets_min": 1051 | |
| }, | |
| { | |
| "epoch": 5.8706070287539935, | |
| "grad_norm": 0.4252831347946701, | |
| "learning_rate": 3.0986008714000703e-06, | |
| "loss": 0.1266, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1176486611366272, | |
| "step": 3675, | |
| "valid_targets_mean": 4929.7, | |
| "valid_targets_min": 2304 | |
| }, | |
| { | |
| "epoch": 5.878594249201278, | |
| "grad_norm": 0.4281154983559537, | |
| "learning_rate": 3.056136302262489e-06, | |
| "loss": 0.1395, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1280544251203537, | |
| "step": 3680, | |
| "valid_targets_mean": 4683.0, | |
| "valid_targets_min": 766 | |
| }, | |
| { | |
| "epoch": 5.886581469648562, | |
| "grad_norm": 0.4491727727952088, | |
| "learning_rate": 3.0139406376278212e-06, | |
| "loss": 0.1277, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1371842324733734, | |
| "step": 3685, | |
| "valid_targets_mean": 4710.4, | |
| "valid_targets_min": 2639 | |
| }, | |
| { | |
| "epoch": 5.894568690095847, | |
| "grad_norm": 0.4278705482057216, | |
| "learning_rate": 2.972014547154671e-06, | |
| "loss": 0.1316, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1220925971865654, | |
| "step": 3690, | |
| "valid_targets_mean": 5169.3, | |
| "valid_targets_min": 3059 | |
| }, | |
| { | |
| "epoch": 5.902555910543131, | |
| "grad_norm": 0.4227843768432032, | |
| "learning_rate": 2.930358696223423e-06, | |
| "loss": 0.1257, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13430209457874298, | |
| "step": 3695, | |
| "valid_targets_mean": 5107.5, | |
| "valid_targets_min": 3469 | |
| }, | |
| { | |
| "epoch": 5.9105431309904155, | |
| "grad_norm": 0.3984550777493105, | |
| "learning_rate": 2.8889737459256695e-06, | |
| "loss": 0.1298, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12288731336593628, | |
| "step": 3700, | |
| "valid_targets_mean": 5928.4, | |
| "valid_targets_min": 2138 | |
| }, | |
| { | |
| "epoch": 5.9185303514377, | |
| "grad_norm": 0.44393618778368915, | |
| "learning_rate": 2.8478603530537285e-06, | |
| "loss": 0.1296, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15052321553230286, | |
| "step": 3705, | |
| "valid_targets_mean": 5530.3, | |
| "valid_targets_min": 2442 | |
| }, | |
| { | |
| "epoch": 5.926517571884984, | |
| "grad_norm": 0.43584878547187356, | |
| "learning_rate": 2.8070191700902194e-06, | |
| "loss": 0.1318, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11622603237628937, | |
| "step": 3710, | |
| "valid_targets_mean": 5041.4, | |
| "valid_targets_min": 2652 | |
| }, | |
| { | |
| "epoch": 5.934504792332269, | |
| "grad_norm": 0.4245456523100115, | |
| "learning_rate": 2.7664508451976903e-06, | |
| "loss": 0.1326, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12009280920028687, | |
| "step": 3715, | |
| "valid_targets_mean": 5665.1, | |
| "valid_targets_min": 2534 | |
| }, | |
| { | |
| "epoch": 5.942492012779553, | |
| "grad_norm": 0.45525403925032903, | |
| "learning_rate": 2.726156022208362e-06, | |
| "loss": 0.1258, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12615250051021576, | |
| "step": 3720, | |
| "valid_targets_mean": 5045.8, | |
| "valid_targets_min": 1735 | |
| }, | |
| { | |
| "epoch": 5.950479233226837, | |
| "grad_norm": 0.4835709648320356, | |
| "learning_rate": 2.6861353406138713e-06, | |
| "loss": 0.1337, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1616552174091339, | |
| "step": 3725, | |
| "valid_targets_mean": 4733.4, | |
| "valid_targets_min": 1895 | |
| }, | |
| { | |
| "epoch": 5.958466453674122, | |
| "grad_norm": 0.45230334079830065, | |
| "learning_rate": 2.646389435555172e-06, | |
| "loss": 0.1355, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16523957252502441, | |
| "step": 3730, | |
| "valid_targets_mean": 4969.8, | |
| "valid_targets_min": 2459 | |
| }, | |
| { | |
| "epoch": 5.966453674121405, | |
| "grad_norm": 0.4485924387968722, | |
| "learning_rate": 2.6069189378124015e-06, | |
| "loss": 0.1279, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12285726517438889, | |
| "step": 3735, | |
| "valid_targets_mean": 5182.8, | |
| "valid_targets_min": 2609 | |
| }, | |
| { | |
| "epoch": 5.97444089456869, | |
| "grad_norm": 0.4133817134086237, | |
| "learning_rate": 2.567724473794908e-06, | |
| "loss": 0.1311, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11647310853004456, | |
| "step": 3740, | |
| "valid_targets_mean": 5128.2, | |
| "valid_targets_min": 1951 | |
| }, | |
| { | |
| "epoch": 5.982428115015974, | |
| "grad_norm": 0.4370931248095585, | |
| "learning_rate": 2.5288066655312914e-06, | |
| "loss": 0.1401, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10781089961528778, | |
| "step": 3745, | |
| "valid_targets_mean": 4750.8, | |
| "valid_targets_min": 2698 | |
| }, | |
| { | |
| "epoch": 5.9904153354632586, | |
| "grad_norm": 0.4139770440675781, | |
| "learning_rate": 2.4901661306595414e-06, | |
| "loss": 0.1292, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13218185305595398, | |
| "step": 3750, | |
| "valid_targets_mean": 5484.0, | |
| "valid_targets_min": 2932 | |
| }, | |
| { | |
| "epoch": 5.998402555910543, | |
| "grad_norm": 0.4326718608578586, | |
| "learning_rate": 2.451803482417234e-06, | |
| "loss": 0.1376, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13455569744110107, | |
| "step": 3755, | |
| "valid_targets_mean": 5015.6, | |
| "valid_targets_min": 772 | |
| }, | |
| { | |
| "epoch": 6.006389776357827, | |
| "grad_norm": 0.5460241672787738, | |
| "learning_rate": 2.413719329631785e-06, | |
| "loss": 0.1325, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13558244705200195, | |
| "step": 3760, | |
| "valid_targets_mean": 5379.9, | |
| "valid_targets_min": 275 | |
| }, | |
| { | |
| "epoch": 6.014376996805112, | |
| "grad_norm": 0.45815263274175894, | |
| "learning_rate": 2.375914276710811e-06, | |
| "loss": 0.1339, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14202500879764557, | |
| "step": 3765, | |
| "valid_targets_mean": 5033.2, | |
| "valid_targets_min": 2205 | |
| }, | |
| { | |
| "epoch": 6.022364217252396, | |
| "grad_norm": 0.4812802636829504, | |
| "learning_rate": 2.338388923632513e-06, | |
| "loss": 0.1245, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13519111275672913, | |
| "step": 3770, | |
| "valid_targets_mean": 5092.1, | |
| "valid_targets_min": 1943 | |
| }, | |
| { | |
| "epoch": 6.0303514376996805, | |
| "grad_norm": 0.41807197895565634, | |
| "learning_rate": 2.3011438659361794e-06, | |
| "loss": 0.138, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13330194354057312, | |
| "step": 3775, | |
| "valid_targets_mean": 5375.8, | |
| "valid_targets_min": 2396 | |
| }, | |
| { | |
| "epoch": 6.038338658146965, | |
| "grad_norm": 0.4253653325419097, | |
| "learning_rate": 2.2641796947127114e-06, | |
| "loss": 0.1201, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10713401436805725, | |
| "step": 3780, | |
| "valid_targets_mean": 4474.6, | |
| "valid_targets_min": 1533 | |
| }, | |
| { | |
| "epoch": 6.046325878594249, | |
| "grad_norm": 0.4396665264148446, | |
| "learning_rate": 2.2274969965952553e-06, | |
| "loss": 0.1291, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12288574874401093, | |
| "step": 3785, | |
| "valid_targets_mean": 4980.4, | |
| "valid_targets_min": 1167 | |
| }, | |
| { | |
| "epoch": 6.054313099041534, | |
| "grad_norm": 0.45323319872441004, | |
| "learning_rate": 2.1910963537498887e-06, | |
| "loss": 0.1229, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13498860597610474, | |
| "step": 3790, | |
| "valid_targets_mean": 4656.8, | |
| "valid_targets_min": 865 | |
| }, | |
| { | |
| "epoch": 6.062300319488818, | |
| "grad_norm": 0.46917552000176416, | |
| "learning_rate": 2.1549783438663872e-06, | |
| "loss": 0.128, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13069258630275726, | |
| "step": 3795, | |
| "valid_targets_mean": 4506.9, | |
| "valid_targets_min": 2171 | |
| }, | |
| { | |
| "epoch": 6.0702875399361025, | |
| "grad_norm": 0.5161681258104249, | |
| "learning_rate": 2.1191435401490534e-06, | |
| "loss": 0.1236, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1369452327489853, | |
| "step": 3800, | |
| "valid_targets_mean": 4470.9, | |
| "valid_targets_min": 2425 | |
| }, | |
| { | |
| "epoch": 6.078274760383387, | |
| "grad_norm": 0.4075234630406859, | |
| "learning_rate": 2.0835925113076062e-06, | |
| "loss": 0.1157, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10623699426651001, | |
| "step": 3805, | |
| "valid_targets_mean": 5392.6, | |
| "valid_targets_min": 2814 | |
| }, | |
| { | |
| "epoch": 6.086261980830671, | |
| "grad_norm": 0.445088891374391, | |
| "learning_rate": 2.0483258215481784e-06, | |
| "loss": 0.1306, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16266600787639618, | |
| "step": 3810, | |
| "valid_targets_mean": 5446.0, | |
| "valid_targets_min": 2279 | |
| }, | |
| { | |
| "epoch": 6.094249201277956, | |
| "grad_norm": 0.43210444561310346, | |
| "learning_rate": 2.0133440305643413e-06, | |
| "loss": 0.118, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12760448455810547, | |
| "step": 3815, | |
| "valid_targets_mean": 5310.2, | |
| "valid_targets_min": 1945 | |
| }, | |
| { | |
| "epoch": 6.102236421725239, | |
| "grad_norm": 0.40439435781857946, | |
| "learning_rate": 1.9786476935282463e-06, | |
| "loss": 0.1329, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12018898874521255, | |
| "step": 3820, | |
| "valid_targets_mean": 5774.5, | |
| "valid_targets_min": 2163 | |
| }, | |
| { | |
| "epoch": 6.110223642172524, | |
| "grad_norm": 0.42516273527603143, | |
| "learning_rate": 1.944237361081782e-06, | |
| "loss": 0.1229, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12302274256944656, | |
| "step": 3825, | |
| "valid_targets_mean": 5523.6, | |
| "valid_targets_min": 2794 | |
| }, | |
| { | |
| "epoch": 6.118210862619808, | |
| "grad_norm": 0.43400398733901474, | |
| "learning_rate": 1.9101135793278746e-06, | |
| "loss": 0.1309, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1362086683511734, | |
| "step": 3830, | |
| "valid_targets_mean": 5282.4, | |
| "valid_targets_min": 3178 | |
| }, | |
| { | |
| "epoch": 6.126198083067092, | |
| "grad_norm": 0.44987633792264775, | |
| "learning_rate": 1.8762768898217732e-06, | |
| "loss": 0.1236, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10965052992105484, | |
| "step": 3835, | |
| "valid_targets_mean": 4636.4, | |
| "valid_targets_min": 2391 | |
| }, | |
| { | |
| "epoch": 6.134185303514377, | |
| "grad_norm": 0.4343409861873823, | |
| "learning_rate": 1.8427278295625006e-06, | |
| "loss": 0.1263, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12411119043827057, | |
| "step": 3840, | |
| "valid_targets_mean": 5069.2, | |
| "valid_targets_min": 1916 | |
| }, | |
| { | |
| "epoch": 6.142172523961661, | |
| "grad_norm": 0.43873354777531515, | |
| "learning_rate": 1.8094669309843161e-06, | |
| "loss": 0.1255, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10989194363355637, | |
| "step": 3845, | |
| "valid_targets_mean": 4884.4, | |
| "valid_targets_min": 655 | |
| }, | |
| { | |
| "epoch": 6.1501597444089455, | |
| "grad_norm": 0.4633346174183801, | |
| "learning_rate": 1.776494721948241e-06, | |
| "loss": 0.1316, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1340833455324173, | |
| "step": 3850, | |
| "valid_targets_mean": 5651.1, | |
| "valid_targets_min": 2898 | |
| }, | |
| { | |
| "epoch": 6.15814696485623, | |
| "grad_norm": 0.427114339635733, | |
| "learning_rate": 1.7438117257337239e-06, | |
| "loss": 0.1181, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.119782954454422, | |
| "step": 3855, | |
| "valid_targets_mean": 4955.4, | |
| "valid_targets_min": 2260 | |
| }, | |
| { | |
| "epoch": 6.166134185303514, | |
| "grad_norm": 0.4196634477811486, | |
| "learning_rate": 1.7114184610302964e-06, | |
| "loss": 0.1115, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10799158364534378, | |
| "step": 3860, | |
| "valid_targets_mean": 5253.5, | |
| "valid_targets_min": 2116 | |
| }, | |
| { | |
| "epoch": 6.174121405750799, | |
| "grad_norm": 0.46204998307303347, | |
| "learning_rate": 1.67931544192937e-06, | |
| "loss": 0.1276, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15535983443260193, | |
| "step": 3865, | |
| "valid_targets_mean": 5263.3, | |
| "valid_targets_min": 748 | |
| }, | |
| { | |
| "epoch": 6.182108626198083, | |
| "grad_norm": 0.4394121104663301, | |
| "learning_rate": 1.6475031779160611e-06, | |
| "loss": 0.1189, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12357381731271744, | |
| "step": 3870, | |
| "valid_targets_mean": 5735.9, | |
| "valid_targets_min": 2900 | |
| }, | |
| { | |
| "epoch": 6.1900958466453675, | |
| "grad_norm": 0.4364702652737178, | |
| "learning_rate": 1.6159821738611192e-06, | |
| "loss": 0.1212, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12687057256698608, | |
| "step": 3875, | |
| "valid_targets_mean": 5862.4, | |
| "valid_targets_min": 3148 | |
| }, | |
| { | |
| "epoch": 6.198083067092652, | |
| "grad_norm": 0.43851879229715685, | |
| "learning_rate": 1.5847529300128827e-06, | |
| "loss": 0.1302, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13917067646980286, | |
| "step": 3880, | |
| "valid_targets_mean": 4974.2, | |
| "valid_targets_min": 726 | |
| }, | |
| { | |
| "epoch": 6.206070287539936, | |
| "grad_norm": 0.4541386352455341, | |
| "learning_rate": 1.5538159419893895e-06, | |
| "loss": 0.1264, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1502765715122223, | |
| "step": 3885, | |
| "valid_targets_mean": 5066.8, | |
| "valid_targets_min": 2980 | |
| }, | |
| { | |
| "epoch": 6.214057507987221, | |
| "grad_norm": 0.44960912003640174, | |
| "learning_rate": 1.5231717007704738e-06, | |
| "loss": 0.1287, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13186979293823242, | |
| "step": 3890, | |
| "valid_targets_mean": 5338.2, | |
| "valid_targets_min": 1838 | |
| }, | |
| { | |
| "epoch": 6.222044728434505, | |
| "grad_norm": 0.4666562128364741, | |
| "learning_rate": 1.4928206926899801e-06, | |
| "loss": 0.1337, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13739527761936188, | |
| "step": 3895, | |
| "valid_targets_mean": 5301.6, | |
| "valid_targets_min": 3537 | |
| }, | |
| { | |
| "epoch": 6.2300319488817895, | |
| "grad_norm": 0.48942259606385624, | |
| "learning_rate": 1.4627633994280599e-06, | |
| "loss": 0.1226, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1267428696155548, | |
| "step": 3900, | |
| "valid_targets_mean": 4616.0, | |
| "valid_targets_min": 1188 | |
| }, | |
| { | |
| "epoch": 6.238019169329074, | |
| "grad_norm": 0.4192740853074209, | |
| "learning_rate": 1.433000298003504e-06, | |
| "loss": 0.1286, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12975162267684937, | |
| "step": 3905, | |
| "valid_targets_mean": 5796.5, | |
| "valid_targets_min": 829 | |
| }, | |
| { | |
| "epoch": 6.246006389776358, | |
| "grad_norm": 0.5044511817012802, | |
| "learning_rate": 1.4035318607662029e-06, | |
| "loss": 0.1271, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10983052849769592, | |
| "step": 3910, | |
| "valid_targets_mean": 5208.8, | |
| "valid_targets_min": 2128 | |
| }, | |
| { | |
| "epoch": 6.253993610223642, | |
| "grad_norm": 0.43769799365232626, | |
| "learning_rate": 1.3743585553896144e-06, | |
| "loss": 0.1273, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11781002581119537, | |
| "step": 3915, | |
| "valid_targets_mean": 5126.4, | |
| "valid_targets_min": 1406 | |
| }, | |
| { | |
| "epoch": 6.261980830670926, | |
| "grad_norm": 0.4439919736525112, | |
| "learning_rate": 1.345480844863376e-06, | |
| "loss": 0.1232, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10559962689876556, | |
| "step": 3920, | |
| "valid_targets_mean": 4546.0, | |
| "valid_targets_min": 1234 | |
| }, | |
| { | |
| "epoch": 6.2699680511182105, | |
| "grad_norm": 0.44001475566889814, | |
| "learning_rate": 1.316899187485925e-06, | |
| "loss": 0.1276, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12955638766288757, | |
| "step": 3925, | |
| "valid_targets_mean": 5041.0, | |
| "valid_targets_min": 2888 | |
| }, | |
| { | |
| "epoch": 6.277955271565495, | |
| "grad_norm": 0.4490867624620665, | |
| "learning_rate": 1.288614036857252e-06, | |
| "loss": 0.125, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12610605359077454, | |
| "step": 3930, | |
| "valid_targets_mean": 4500.5, | |
| "valid_targets_min": 1600 | |
| }, | |
| { | |
| "epoch": 6.285942492012779, | |
| "grad_norm": 0.4573882062510721, | |
| "learning_rate": 1.260625841871692e-06, | |
| "loss": 0.1245, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12089214473962784, | |
| "step": 3935, | |
| "valid_targets_mean": 5326.1, | |
| "valid_targets_min": 1714 | |
| }, | |
| { | |
| "epoch": 6.293929712460064, | |
| "grad_norm": 0.460320180700977, | |
| "learning_rate": 1.2329350467107925e-06, | |
| "loss": 0.1205, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11667491495609283, | |
| "step": 3940, | |
| "valid_targets_mean": 5346.1, | |
| "valid_targets_min": 944 | |
| }, | |
| { | |
| "epoch": 6.301916932907348, | |
| "grad_norm": 0.43256058980425616, | |
| "learning_rate": 1.2055420908362781e-06, | |
| "loss": 0.1149, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10495949536561966, | |
| "step": 3945, | |
| "valid_targets_mean": 5445.4, | |
| "valid_targets_min": 336 | |
| }, | |
| { | |
| "epoch": 6.3099041533546325, | |
| "grad_norm": 0.44051620478323134, | |
| "learning_rate": 1.1784474089830612e-06, | |
| "loss": 0.121, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13951703906059265, | |
| "step": 3950, | |
| "valid_targets_mean": 5139.0, | |
| "valid_targets_min": 2663 | |
| }, | |
| { | |
| "epoch": 6.317891373801917, | |
| "grad_norm": 0.43492039544706806, | |
| "learning_rate": 1.1516514311523607e-06, | |
| "loss": 0.129, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14872542023658752, | |
| "step": 3955, | |
| "valid_targets_mean": 5421.0, | |
| "valid_targets_min": 1835 | |
| }, | |
| { | |
| "epoch": 6.325878594249201, | |
| "grad_norm": 0.4069754142819716, | |
| "learning_rate": 1.1251545826048593e-06, | |
| "loss": 0.1473, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1252405047416687, | |
| "step": 3960, | |
| "valid_targets_mean": 5784.8, | |
| "valid_targets_min": 3965 | |
| }, | |
| { | |
| "epoch": 6.333865814696486, | |
| "grad_norm": 0.44883578106716204, | |
| "learning_rate": 1.098957283853972e-06, | |
| "loss": 0.126, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11359164118766785, | |
| "step": 3965, | |
| "valid_targets_mean": 4839.5, | |
| "valid_targets_min": 1520 | |
| }, | |
| { | |
| "epoch": 6.34185303514377, | |
| "grad_norm": 0.43067946431963217, | |
| "learning_rate": 1.0730599506591476e-06, | |
| "loss": 0.1341, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11182200908660889, | |
| "step": 3970, | |
| "valid_targets_mean": 4953.1, | |
| "valid_targets_min": 2350 | |
| }, | |
| { | |
| "epoch": 6.3498402555910545, | |
| "grad_norm": 0.42868136031067866, | |
| "learning_rate": 1.0474629940192994e-06, | |
| "loss": 0.1242, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11280225962400436, | |
| "step": 3975, | |
| "valid_targets_mean": 5008.3, | |
| "valid_targets_min": 2005 | |
| }, | |
| { | |
| "epoch": 6.357827476038339, | |
| "grad_norm": 0.46038734388380154, | |
| "learning_rate": 1.0221668201662726e-06, | |
| "loss": 0.1296, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11624015867710114, | |
| "step": 3980, | |
| "valid_targets_mean": 5058.4, | |
| "valid_targets_min": 2599 | |
| }, | |
| { | |
| "epoch": 6.365814696485623, | |
| "grad_norm": 0.42039366706539916, | |
| "learning_rate": 9.971718305583767e-07, | |
| "loss": 0.1167, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10677272081375122, | |
| "step": 3985, | |
| "valid_targets_mean": 5019.1, | |
| "valid_targets_min": 1599 | |
| }, | |
| { | |
| "epoch": 6.373801916932908, | |
| "grad_norm": 0.41108665824144747, | |
| "learning_rate": 9.724784218740524e-07, | |
| "loss": 0.1285, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1203266903758049, | |
| "step": 3990, | |
| "valid_targets_mean": 5354.8, | |
| "valid_targets_min": 3058 | |
| }, | |
| { | |
| "epoch": 6.381789137380192, | |
| "grad_norm": 0.460679377059388, | |
| "learning_rate": 9.480869860055364e-07, | |
| "loss": 0.1299, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14629745483398438, | |
| "step": 3995, | |
| "valid_targets_mean": 5655.1, | |
| "valid_targets_min": 2461 | |
| }, | |
| { | |
| "epoch": 6.389776357827476, | |
| "grad_norm": 0.5037933215304593, | |
| "learning_rate": 9.239979100526763e-07, | |
| "loss": 0.1315, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11931949108839035, | |
| "step": 4000, | |
| "valid_targets_mean": 4506.9, | |
| "valid_targets_min": 2251 | |
| }, | |
| { | |
| "epoch": 6.397763578274761, | |
| "grad_norm": 0.42901480635849415, | |
| "learning_rate": 9.002115763167585e-07, | |
| "loss": 0.1346, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12577411532402039, | |
| "step": 4005, | |
| "valid_targets_mean": 5733.3, | |
| "valid_targets_min": 2281 | |
| }, | |
| { | |
| "epoch": 6.405750798722044, | |
| "grad_norm": 0.49853715959044076, | |
| "learning_rate": 8.7672836229447e-07, | |
| "loss": 0.1274, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13035975396633148, | |
| "step": 4010, | |
| "valid_targets_mean": 4618.9, | |
| "valid_targets_min": 966 | |
| }, | |
| { | |
| "epoch": 6.413738019169329, | |
| "grad_norm": 0.47970641363612365, | |
| "learning_rate": 8.535486406718684e-07, | |
| "loss": 0.1279, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1349407583475113, | |
| "step": 4015, | |
| "valid_targets_mean": 5299.3, | |
| "valid_targets_min": 1770 | |
| }, | |
| { | |
| "epoch": 6.421725239616613, | |
| "grad_norm": 0.4372706109561565, | |
| "learning_rate": 8.306727793185132e-07, | |
| "loss": 0.126, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11875171959400177, | |
| "step": 4020, | |
| "valid_targets_mean": 5256.9, | |
| "valid_targets_min": 1861 | |
| }, | |
| { | |
| "epoch": 6.4297124600638975, | |
| "grad_norm": 0.4348504836224484, | |
| "learning_rate": 8.081011412815965e-07, | |
| "loss": 0.1193, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12242938578128815, | |
| "step": 4025, | |
| "valid_targets_mean": 4903.8, | |
| "valid_targets_min": 2566 | |
| }, | |
| { | |
| "epoch": 6.437699680511182, | |
| "grad_norm": 0.4740085970492133, | |
| "learning_rate": 7.858340847801815e-07, | |
| "loss": 0.1208, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.126695454120636, | |
| "step": 4030, | |
| "valid_targets_mean": 4449.8, | |
| "valid_targets_min": 2033 | |
| }, | |
| { | |
| "epoch": 6.445686900958466, | |
| "grad_norm": 0.44618893769751705, | |
| "learning_rate": 7.638719631995406e-07, | |
| "loss": 0.1294, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14740081131458282, | |
| "step": 4035, | |
| "valid_targets_mean": 5047.9, | |
| "valid_targets_min": 1987 | |
| }, | |
| { | |
| "epoch": 6.453674121405751, | |
| "grad_norm": 0.44815831497045167, | |
| "learning_rate": 7.422151250855214e-07, | |
| "loss": 0.1263, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11599256098270416, | |
| "step": 4040, | |
| "valid_targets_mean": 4892.8, | |
| "valid_targets_min": 2166 | |
| }, | |
| { | |
| "epoch": 6.461661341853035, | |
| "grad_norm": 0.4113264348058772, | |
| "learning_rate": 7.208639141390295e-07, | |
| "loss": 0.1322, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10958003997802734, | |
| "step": 4045, | |
| "valid_targets_mean": 5203.8, | |
| "valid_targets_min": 2281 | |
| }, | |
| { | |
| "epoch": 6.4696485623003195, | |
| "grad_norm": 0.4753581271252471, | |
| "learning_rate": 6.998186692105657e-07, | |
| "loss": 0.1256, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.136691153049469, | |
| "step": 4050, | |
| "valid_targets_mean": 4522.3, | |
| "valid_targets_min": 2014 | |
| }, | |
| { | |
| "epoch": 6.477635782747604, | |
| "grad_norm": 0.43896515321461765, | |
| "learning_rate": 6.790797242948644e-07, | |
| "loss": 0.1234, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11603130400180817, | |
| "step": 4055, | |
| "valid_targets_mean": 4522.9, | |
| "valid_targets_min": 882 | |
| }, | |
| { | |
| "epoch": 6.485623003194888, | |
| "grad_norm": 0.40815258510746033, | |
| "learning_rate": 6.5864740852557e-07, | |
| "loss": 0.1264, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.09882775694131851, | |
| "step": 4060, | |
| "valid_targets_mean": 5323.5, | |
| "valid_targets_min": 2341 | |
| }, | |
| { | |
| "epoch": 6.493610223642173, | |
| "grad_norm": 0.45612748381451096, | |
| "learning_rate": 6.385220461700248e-07, | |
| "loss": 0.1342, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12171460688114166, | |
| "step": 4065, | |
| "valid_targets_mean": 4586.4, | |
| "valid_targets_min": 2863 | |
| }, | |
| { | |
| "epoch": 6.501597444089457, | |
| "grad_norm": 0.43581879450562144, | |
| "learning_rate": 6.187039566241337e-07, | |
| "loss": 0.1271, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10705038905143738, | |
| "step": 4070, | |
| "valid_targets_mean": 5215.1, | |
| "valid_targets_min": 2975 | |
| }, | |
| { | |
| "epoch": 6.5095846645367414, | |
| "grad_norm": 0.4138256432569784, | |
| "learning_rate": 5.99193454407272e-07, | |
| "loss": 0.129, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1212131455540657, | |
| "step": 4075, | |
| "valid_targets_mean": 5568.9, | |
| "valid_targets_min": 3698 | |
| }, | |
| { | |
| "epoch": 6.517571884984026, | |
| "grad_norm": 0.428726314271063, | |
| "learning_rate": 5.799908491573148e-07, | |
| "loss": 0.135, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12733975052833557, | |
| "step": 4080, | |
| "valid_targets_mean": 5212.7, | |
| "valid_targets_min": 3405 | |
| }, | |
| { | |
| "epoch": 6.52555910543131, | |
| "grad_norm": 0.4020835335100901, | |
| "learning_rate": 5.610964456257107e-07, | |
| "loss": 0.1276, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11473269015550613, | |
| "step": 4085, | |
| "valid_targets_mean": 5244.9, | |
| "valid_targets_min": 2164 | |
| }, | |
| { | |
| "epoch": 6.533546325878595, | |
| "grad_norm": 0.4033229463927491, | |
| "learning_rate": 5.425105436726496e-07, | |
| "loss": 0.1233, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10510792583227158, | |
| "step": 4090, | |
| "valid_targets_mean": 5308.1, | |
| "valid_targets_min": 3010 | |
| }, | |
| { | |
| "epoch": 6.541533546325878, | |
| "grad_norm": 0.44044531081284927, | |
| "learning_rate": 5.242334382623004e-07, | |
| "loss": 0.1379, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12793225049972534, | |
| "step": 4095, | |
| "valid_targets_mean": 5412.9, | |
| "valid_targets_min": 2178 | |
| }, | |
| { | |
| "epoch": 6.549520766773163, | |
| "grad_norm": 0.44074657000900863, | |
| "learning_rate": 5.062654194581429e-07, | |
| "loss": 0.1276, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11548326909542084, | |
| "step": 4100, | |
| "valid_targets_mean": 5007.6, | |
| "valid_targets_min": 2544 | |
| }, | |
| { | |
| "epoch": 6.557507987220447, | |
| "grad_norm": 0.46972005517840504, | |
| "learning_rate": 4.88606772418343e-07, | |
| "loss": 0.13, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11469545215368271, | |
| "step": 4105, | |
| "valid_targets_mean": 4550.8, | |
| "valid_targets_min": 2013 | |
| }, | |
| { | |
| "epoch": 6.565495207667731, | |
| "grad_norm": 0.49127287018444826, | |
| "learning_rate": 4.7125777739123857e-07, | |
| "loss": 0.1174, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11164946109056473, | |
| "step": 4110, | |
| "valid_targets_mean": 3931.2, | |
| "valid_targets_min": 1109 | |
| }, | |
| { | |
| "epoch": 6.573482428115016, | |
| "grad_norm": 0.4668107707151476, | |
| "learning_rate": 4.54218709710903e-07, | |
| "loss": 0.1365, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13199590146541595, | |
| "step": 4115, | |
| "valid_targets_mean": 4663.8, | |
| "valid_targets_min": 618 | |
| }, | |
| { | |
| "epoch": 6.5814696485623, | |
| "grad_norm": 0.44382795252910734, | |
| "learning_rate": 4.374898397927507e-07, | |
| "loss": 0.1262, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12120091170072556, | |
| "step": 4120, | |
| "valid_targets_mean": 4897.0, | |
| "valid_targets_min": 2203 | |
| }, | |
| { | |
| "epoch": 6.5894568690095845, | |
| "grad_norm": 0.4300852158838798, | |
| "learning_rate": 4.210714331292698e-07, | |
| "loss": 0.1306, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12831351161003113, | |
| "step": 4125, | |
| "valid_targets_mean": 5346.8, | |
| "valid_targets_min": 2949 | |
| }, | |
| { | |
| "epoch": 6.597444089456869, | |
| "grad_norm": 0.5192194474971673, | |
| "learning_rate": 4.049637502857895e-07, | |
| "loss": 0.1159, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1173935979604721, | |
| "step": 4130, | |
| "valid_targets_mean": 5576.4, | |
| "valid_targets_min": 2912 | |
| }, | |
| { | |
| "epoch": 6.605431309904153, | |
| "grad_norm": 0.5442613394742007, | |
| "learning_rate": 3.8916704689635707e-07, | |
| "loss": 0.1308, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1252613365650177, | |
| "step": 4135, | |
| "valid_targets_mean": 4412.6, | |
| "valid_targets_min": 1744 | |
| }, | |
| { | |
| "epoch": 6.613418530351438, | |
| "grad_norm": 0.4710900621176652, | |
| "learning_rate": 3.736815736596766e-07, | |
| "loss": 0.1259, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13020065426826477, | |
| "step": 4140, | |
| "valid_targets_mean": 5045.9, | |
| "valid_targets_min": 2691 | |
| }, | |
| { | |
| "epoch": 6.621405750798722, | |
| "grad_norm": 0.4441003248760054, | |
| "learning_rate": 3.5850757633513424e-07, | |
| "loss": 0.1282, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12181347608566284, | |
| "step": 4145, | |
| "valid_targets_mean": 5066.2, | |
| "valid_targets_min": 2915 | |
| }, | |
| { | |
| "epoch": 6.6293929712460065, | |
| "grad_norm": 0.4375025143828666, | |
| "learning_rate": 3.4364529573888803e-07, | |
| "loss": 0.1308, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15717995166778564, | |
| "step": 4150, | |
| "valid_targets_mean": 5708.8, | |
| "valid_targets_min": 2743 | |
| }, | |
| { | |
| "epoch": 6.637380191693291, | |
| "grad_norm": 0.46300408010048055, | |
| "learning_rate": 3.2909496774005344e-07, | |
| "loss": 0.1336, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1467522531747818, | |
| "step": 4155, | |
| "valid_targets_mean": 5205.4, | |
| "valid_targets_min": 2501 | |
| }, | |
| { | |
| "epoch": 6.645367412140575, | |
| "grad_norm": 0.41809833193188717, | |
| "learning_rate": 3.14856823256966e-07, | |
| "loss": 0.1307, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12621453404426575, | |
| "step": 4160, | |
| "valid_targets_mean": 5243.4, | |
| "valid_targets_min": 1712 | |
| }, | |
| { | |
| "epoch": 6.65335463258786, | |
| "grad_norm": 0.44709400177245917, | |
| "learning_rate": 3.009310882534999e-07, | |
| "loss": 0.1224, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11646348983049393, | |
| "step": 4165, | |
| "valid_targets_mean": 5243.4, | |
| "valid_targets_min": 2857 | |
| }, | |
| { | |
| "epoch": 6.661341853035144, | |
| "grad_norm": 0.47958560127019007, | |
| "learning_rate": 2.8731798373550004e-07, | |
| "loss": 0.1276, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13368624448776245, | |
| "step": 4170, | |
| "valid_targets_mean": 5001.4, | |
| "valid_targets_min": 1928 | |
| }, | |
| { | |
| "epoch": 6.669329073482428, | |
| "grad_norm": 0.4711690703941965, | |
| "learning_rate": 2.740177257472576e-07, | |
| "loss": 0.1239, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1244480311870575, | |
| "step": 4175, | |
| "valid_targets_mean": 4708.8, | |
| "valid_targets_min": 1570 | |
| }, | |
| { | |
| "epoch": 6.677316293929713, | |
| "grad_norm": 0.44654010468455774, | |
| "learning_rate": 2.6103052536810226e-07, | |
| "loss": 0.1179, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1290716677904129, | |
| "step": 4180, | |
| "valid_targets_mean": 4989.8, | |
| "valid_targets_min": 2551 | |
| }, | |
| { | |
| "epoch": 6.685303514376997, | |
| "grad_norm": 0.4607336478048513, | |
| "learning_rate": 2.4835658870902226e-07, | |
| "loss": 0.1201, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10677691549062729, | |
| "step": 4185, | |
| "valid_targets_mean": 4670.4, | |
| "valid_targets_min": 1438 | |
| }, | |
| { | |
| "epoch": 6.693290734824281, | |
| "grad_norm": 0.40814159756552487, | |
| "learning_rate": 2.3599611690943158e-07, | |
| "loss": 0.1291, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12436793744564056, | |
| "step": 4190, | |
| "valid_targets_mean": 5735.2, | |
| "valid_targets_min": 3568 | |
| }, | |
| { | |
| "epoch": 6.701277955271565, | |
| "grad_norm": 0.4324558117695401, | |
| "learning_rate": 2.2394930613393927e-07, | |
| "loss": 0.1257, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13884522020816803, | |
| "step": 4195, | |
| "valid_targets_mean": 5794.0, | |
| "valid_targets_min": 2887 | |
| }, | |
| { | |
| "epoch": 6.7092651757188495, | |
| "grad_norm": 0.44929256467791345, | |
| "learning_rate": 2.122163475692629e-07, | |
| "loss": 0.1311, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12921413779258728, | |
| "step": 4200, | |
| "valid_targets_mean": 5007.8, | |
| "valid_targets_min": 3512 | |
| }, | |
| { | |
| "epoch": 6.717252396166134, | |
| "grad_norm": 0.46290416248413574, | |
| "learning_rate": 2.0079742742118878e-07, | |
| "loss": 0.116, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1145065575838089, | |
| "step": 4205, | |
| "valid_targets_mean": 5079.2, | |
| "valid_targets_min": 2893 | |
| }, | |
| { | |
| "epoch": 6.725239616613418, | |
| "grad_norm": 0.43545148648922827, | |
| "learning_rate": 1.8969272691160334e-07, | |
| "loss": 0.1243, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13240908086299896, | |
| "step": 4210, | |
| "valid_targets_mean": 5030.4, | |
| "valid_targets_min": 601 | |
| }, | |
| { | |
| "epoch": 6.733226837060703, | |
| "grad_norm": 0.419155697642424, | |
| "learning_rate": 1.789024222756397e-07, | |
| "loss": 0.1385, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11393342912197113, | |
| "step": 4215, | |
| "valid_targets_mean": 5162.2, | |
| "valid_targets_min": 3070 | |
| }, | |
| { | |
| "epoch": 6.741214057507987, | |
| "grad_norm": 0.4744305970205998, | |
| "learning_rate": 1.6842668475885782e-07, | |
| "loss": 0.1255, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1319217085838318, | |
| "step": 4220, | |
| "valid_targets_mean": 5766.8, | |
| "valid_targets_min": 3464 | |
| }, | |
| { | |
| "epoch": 6.7492012779552715, | |
| "grad_norm": 0.4828292996584737, | |
| "learning_rate": 1.582656806145444e-07, | |
| "loss": 0.1304, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1557837426662445, | |
| "step": 4225, | |
| "valid_targets_mean": 5423.4, | |
| "valid_targets_min": 2528 | |
| }, | |
| { | |
| "epoch": 6.757188498402556, | |
| "grad_norm": 0.4182357988469141, | |
| "learning_rate": 1.4841957110106388e-07, | |
| "loss": 0.1263, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13060078024864197, | |
| "step": 4230, | |
| "valid_targets_mean": 5163.9, | |
| "valid_targets_min": 2830 | |
| }, | |
| { | |
| "epoch": 6.76517571884984, | |
| "grad_norm": 0.4107576102937048, | |
| "learning_rate": 1.388885124793049e-07, | |
| "loss": 0.1274, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.123427614569664, | |
| "step": 4235, | |
| "valid_targets_mean": 5327.9, | |
| "valid_targets_min": 438 | |
| }, | |
| { | |
| "epoch": 6.773162939297125, | |
| "grad_norm": 0.4829003992594771, | |
| "learning_rate": 1.2967265601019573e-07, | |
| "loss": 0.1305, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14884167909622192, | |
| "step": 4240, | |
| "valid_targets_mean": 4674.9, | |
| "valid_targets_min": 2476 | |
| }, | |
| { | |
| "epoch": 6.781150159744409, | |
| "grad_norm": 0.4551560271428259, | |
| "learning_rate": 1.207721479523105e-07, | |
| "loss": 0.1326, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14202353358268738, | |
| "step": 4245, | |
| "valid_targets_mean": 5272.5, | |
| "valid_targets_min": 1914 | |
| }, | |
| { | |
| "epoch": 6.789137380191693, | |
| "grad_norm": 0.4785568483806696, | |
| "learning_rate": 1.1218712955954003e-07, | |
| "loss": 0.119, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13617315888404846, | |
| "step": 4250, | |
| "valid_targets_mean": 4645.4, | |
| "valid_targets_min": 1815 | |
| }, | |
| { | |
| "epoch": 6.797124600638978, | |
| "grad_norm": 0.421600006237604, | |
| "learning_rate": 1.0391773707885578e-07, | |
| "loss": 0.1238, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.123545341193676, | |
| "step": 4255, | |
| "valid_targets_mean": 5768.4, | |
| "valid_targets_min": 1951 | |
| }, | |
| { | |
| "epoch": 6.805111821086262, | |
| "grad_norm": 0.4891057855574066, | |
| "learning_rate": 9.596410174814497e-08, | |
| "loss": 0.1207, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11968959867954254, | |
| "step": 4260, | |
| "valid_targets_mean": 4024.6, | |
| "valid_targets_min": 2162 | |
| }, | |
| { | |
| "epoch": 6.813099041533547, | |
| "grad_norm": 0.42245162265441527, | |
| "learning_rate": 8.83263497941278e-08, | |
| "loss": 0.1161, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10335589200258255, | |
| "step": 4265, | |
| "valid_targets_mean": 5149.1, | |
| "valid_targets_min": 2548 | |
| }, | |
| { | |
| "epoch": 6.821086261980831, | |
| "grad_norm": 0.4671241584555409, | |
| "learning_rate": 8.10046024303568e-08, | |
| "loss": 0.1178, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1074795052409172, | |
| "step": 4270, | |
| "valid_targets_mean": 4250.1, | |
| "valid_targets_min": 1519 | |
| }, | |
| { | |
| "epoch": 6.8290734824281145, | |
| "grad_norm": 0.43561092040757404, | |
| "learning_rate": 7.399897585528726e-08, | |
| "loss": 0.1295, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11564867943525314, | |
| "step": 4275, | |
| "valid_targets_mean": 4972.9, | |
| "valid_targets_min": 1407 | |
| }, | |
| { | |
| "epoch": 6.8370607028754, | |
| "grad_norm": 0.4115609138657699, | |
| "learning_rate": 6.730958125044319e-08, | |
| "loss": 0.1282, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11483149230480194, | |
| "step": 4280, | |
| "valid_targets_mean": 5207.3, | |
| "valid_targets_min": 2197 | |
| }, | |
| { | |
| "epoch": 6.845047923322683, | |
| "grad_norm": 0.46699280764135903, | |
| "learning_rate": 6.093652477864309e-08, | |
| "loss": 0.1161, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11931362748146057, | |
| "step": 4285, | |
| "valid_targets_mean": 4871.9, | |
| "valid_targets_min": 1608 | |
| }, | |
| { | |
| "epoch": 6.853035143769968, | |
| "grad_norm": 0.45645715118430497, | |
| "learning_rate": 5.4879907582316986e-08, | |
| "loss": 0.1374, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12514854967594147, | |
| "step": 4290, | |
| "valid_targets_mean": 4269.4, | |
| "valid_targets_min": 1736 | |
| }, | |
| { | |
| "epoch": 6.861022364217252, | |
| "grad_norm": 0.4291023452764324, | |
| "learning_rate": 4.913982578190535e-08, | |
| "loss": 0.1234, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11251398175954819, | |
| "step": 4295, | |
| "valid_targets_mean": 5025.5, | |
| "valid_targets_min": 2160 | |
| }, | |
| { | |
| "epoch": 6.8690095846645365, | |
| "grad_norm": 0.49394093785345944, | |
| "learning_rate": 4.3716370474331527e-08, | |
| "loss": 0.1326, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13623955845832825, | |
| "step": 4300, | |
| "valid_targets_mean": 4479.9, | |
| "valid_targets_min": 744 | |
| }, | |
| { | |
| "epoch": 6.876996805111821, | |
| "grad_norm": 0.479353031742284, | |
| "learning_rate": 3.8609627731558405e-08, | |
| "loss": 0.1326, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14167432487010956, | |
| "step": 4305, | |
| "valid_targets_mean": 4416.4, | |
| "valid_targets_min": 1016 | |
| }, | |
| { | |
| "epoch": 6.884984025559105, | |
| "grad_norm": 0.44454957776825904, | |
| "learning_rate": 3.381967859920954e-08, | |
| "loss": 0.1223, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11437994241714478, | |
| "step": 4310, | |
| "valid_targets_mean": 4810.2, | |
| "valid_targets_min": 700 | |
| }, | |
| { | |
| "epoch": 6.89297124600639, | |
| "grad_norm": 0.38364400111284386, | |
| "learning_rate": 2.9346599095305685e-08, | |
| "loss": 0.1364, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15792012214660645, | |
| "step": 4315, | |
| "valid_targets_mean": 6922.1, | |
| "valid_targets_min": 2436 | |
| }, | |
| { | |
| "epoch": 6.900958466453674, | |
| "grad_norm": 0.43111371911991647, | |
| "learning_rate": 2.5190460209039146e-08, | |
| "loss": 0.1333, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11298054456710815, | |
| "step": 4320, | |
| "valid_targets_mean": 5096.0, | |
| "valid_targets_min": 2605 | |
| }, | |
| { | |
| "epoch": 6.9089456869009584, | |
| "grad_norm": 0.447483466824999, | |
| "learning_rate": 2.1351327899656883e-08, | |
| "loss": 0.1277, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1575692594051361, | |
| "step": 4325, | |
| "valid_targets_mean": 5471.5, | |
| "valid_targets_min": 2446 | |
| }, | |
| { | |
| "epoch": 6.916932907348243, | |
| "grad_norm": 0.5414456690357554, | |
| "learning_rate": 1.782926309540578e-08, | |
| "loss": 0.12, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12004803866147995, | |
| "step": 4330, | |
| "valid_targets_mean": 5218.1, | |
| "valid_targets_min": 2319 | |
| }, | |
| { | |
| "epoch": 6.924920127795527, | |
| "grad_norm": 0.44717716422719683, | |
| "learning_rate": 1.462432169257344e-08, | |
| "loss": 0.1234, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11098746955394745, | |
| "step": 4335, | |
| "valid_targets_mean": 5052.2, | |
| "valid_targets_min": 2343 | |
| }, | |
| { | |
| "epoch": 6.932907348242812, | |
| "grad_norm": 0.422353954550692, | |
| "learning_rate": 1.1736554554604429e-08, | |
| "loss": 0.1209, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12036773562431335, | |
| "step": 4340, | |
| "valid_targets_mean": 5242.9, | |
| "valid_targets_min": 1956 | |
| }, | |
| { | |
| "epoch": 6.940894568690096, | |
| "grad_norm": 0.4617476563528072, | |
| "learning_rate": 9.166007511274278e-09, | |
| "loss": 0.1293, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11139798164367676, | |
| "step": 4345, | |
| "valid_targets_mean": 4558.4, | |
| "valid_targets_min": 447 | |
| }, | |
| { | |
| "epoch": 6.94888178913738, | |
| "grad_norm": 0.442867489372069, | |
| "learning_rate": 6.912721357985597e-09, | |
| "loss": 0.1323, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13617973029613495, | |
| "step": 4350, | |
| "valid_targets_mean": 5176.3, | |
| "valid_targets_min": 3292 | |
| }, | |
| { | |
| "epoch": 6.956869009584665, | |
| "grad_norm": 0.45193650974824, | |
| "learning_rate": 4.976731855104166e-09, | |
| "loss": 0.1264, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14197497069835663, | |
| "step": 4355, | |
| "valid_targets_mean": 4814.2, | |
| "valid_targets_min": 903 | |
| }, | |
| { | |
| "epoch": 6.964856230031949, | |
| "grad_norm": 0.5738394349043882, | |
| "learning_rate": 3.3580697274016028e-09, | |
| "loss": 0.1278, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14240266382694244, | |
| "step": 4360, | |
| "valid_targets_mean": 4527.2, | |
| "valid_targets_min": 1976 | |
| }, | |
| { | |
| "epoch": 6.972843450479234, | |
| "grad_norm": 0.47624631751909785, | |
| "learning_rate": 2.056760663555757e-09, | |
| "loss": 0.1265, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12048348784446716, | |
| "step": 4365, | |
| "valid_targets_mean": 4020.9, | |
| "valid_targets_min": 1527 | |
| }, | |
| { | |
| "epoch": 6.980830670926517, | |
| "grad_norm": 0.42331464845965033, | |
| "learning_rate": 1.0728253157599178e-09, | |
| "loss": 0.1283, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1154799610376358, | |
| "step": 4370, | |
| "valid_targets_mean": 5335.5, | |
| "valid_targets_min": 3323 | |
| }, | |
| { | |
| "epoch": 6.988817891373802, | |
| "grad_norm": 0.5003444006546414, | |
| "learning_rate": 4.062792993786424e-10, | |
| "loss": 0.1309, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15271781384944916, | |
| "step": 4375, | |
| "valid_targets_mean": 5267.2, | |
| "valid_targets_min": 3040 | |
| }, | |
| { | |
| "epoch": 6.996805111821086, | |
| "grad_norm": 0.4197559917493364, | |
| "learning_rate": 5.7133192707947705e-11, | |
| "loss": 0.1268, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10710293054580688, | |
| "step": 4380, | |
| "valid_targets_mean": 4864.9, | |
| "valid_targets_min": 2459 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1322430968284607, | |
| "step": 4382, | |
| "total_flos": 1613658688454656.0, | |
| "train_loss": 0.16677221175349505, | |
| "train_runtime": 22726.4582, | |
| "train_samples_per_second": 3.081, | |
| "train_steps_per_second": 0.193, | |
| "valid_targets_mean": 4677.1, | |
| "valid_targets_min": 1685 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 4382, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 7, | |
| "save_steps": 1500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1613658688454656.0, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |