Text Generation
Transformers
Safetensors
qwen3
llama-factory
full
Generated from Trainer
conversational
text-generation-inference
Instructions to use DCAgent/a1-taskmaster2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use DCAgent/a1-taskmaster2 with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="DCAgent/a1-taskmaster2") messages = [ {"role": "user", "content": "Who are you?"}, ] pipe(messages)# Load model directly from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("DCAgent/a1-taskmaster2") model = AutoModelForCausalLM.from_pretrained("DCAgent/a1-taskmaster2", 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-taskmaster2 with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "DCAgent/a1-taskmaster2" # 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-taskmaster2", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/DCAgent/a1-taskmaster2
- SGLang
How to use DCAgent/a1-taskmaster2 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-taskmaster2" \ --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-taskmaster2", "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-taskmaster2" \ --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-taskmaster2", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }' - Docker Model Runner
How to use DCAgent/a1-taskmaster2 with Docker Model Runner:
docker model run hf.co/DCAgent/a1-taskmaster2
| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 7.0, | |
| "eval_steps": 500, | |
| "global_step": 3913, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.008944543828264758, | |
| "grad_norm": 21.606822638718377, | |
| "learning_rate": 4.0816326530612243e-07, | |
| "loss": 0.9157, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.9071354866027832, | |
| "step": 5, | |
| "valid_targets_mean": 4165.1, | |
| "valid_targets_min": 1481 | |
| }, | |
| { | |
| "epoch": 0.017889087656529516, | |
| "grad_norm": 19.04995792657104, | |
| "learning_rate": 9.183673469387756e-07, | |
| "loss": 0.9249, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.8876984119415283, | |
| "step": 10, | |
| "valid_targets_mean": 3925.7, | |
| "valid_targets_min": 1499 | |
| }, | |
| { | |
| "epoch": 0.026833631484794274, | |
| "grad_norm": 19.84564044371412, | |
| "learning_rate": 1.4285714285714286e-06, | |
| "loss": 0.8728, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.8913623690605164, | |
| "step": 15, | |
| "valid_targets_mean": 4151.1, | |
| "valid_targets_min": 2060 | |
| }, | |
| { | |
| "epoch": 0.03577817531305903, | |
| "grad_norm": 12.750163023484031, | |
| "learning_rate": 1.938775510204082e-06, | |
| "loss": 0.8513, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.8227970600128174, | |
| "step": 20, | |
| "valid_targets_mean": 3907.8, | |
| "valid_targets_min": 1805 | |
| }, | |
| { | |
| "epoch": 0.044722719141323794, | |
| "grad_norm": 7.2429705482367055, | |
| "learning_rate": 2.4489795918367347e-06, | |
| "loss": 0.7962, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.7579451203346252, | |
| "step": 25, | |
| "valid_targets_mean": 4374.8, | |
| "valid_targets_min": 538 | |
| }, | |
| { | |
| "epoch": 0.05366726296958855, | |
| "grad_norm": 4.437835296542112, | |
| "learning_rate": 2.959183673469388e-06, | |
| "loss": 0.7424, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.7099573612213135, | |
| "step": 30, | |
| "valid_targets_mean": 3956.7, | |
| "valid_targets_min": 954 | |
| }, | |
| { | |
| "epoch": 0.0626118067978533, | |
| "grad_norm": 2.7488646562577306, | |
| "learning_rate": 3.469387755102041e-06, | |
| "loss": 0.6793, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.6850540041923523, | |
| "step": 35, | |
| "valid_targets_mean": 4000.8, | |
| "valid_targets_min": 579 | |
| }, | |
| { | |
| "epoch": 0.07155635062611806, | |
| "grad_norm": 2.1085770347953767, | |
| "learning_rate": 3.979591836734694e-06, | |
| "loss": 0.6197, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.6641137003898621, | |
| "step": 40, | |
| "valid_targets_mean": 3597.0, | |
| "valid_targets_min": 1574 | |
| }, | |
| { | |
| "epoch": 0.08050089445438283, | |
| "grad_norm": 1.6255237321311984, | |
| "learning_rate": 4.489795918367348e-06, | |
| "loss": 0.6116, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.6154483556747437, | |
| "step": 45, | |
| "valid_targets_mean": 3623.8, | |
| "valid_targets_min": 850 | |
| }, | |
| { | |
| "epoch": 0.08944543828264759, | |
| "grad_norm": 1.2082356599373048, | |
| "learning_rate": 5e-06, | |
| "loss": 0.6101, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.5458171367645264, | |
| "step": 50, | |
| "valid_targets_mean": 3971.1, | |
| "valid_targets_min": 627 | |
| }, | |
| { | |
| "epoch": 0.09838998211091235, | |
| "grad_norm": 1.0504686350094798, | |
| "learning_rate": 5.510204081632653e-06, | |
| "loss": 0.5949, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.573982834815979, | |
| "step": 55, | |
| "valid_targets_mean": 4307.6, | |
| "valid_targets_min": 1734 | |
| }, | |
| { | |
| "epoch": 0.1073345259391771, | |
| "grad_norm": 0.9226116061847508, | |
| "learning_rate": 6.020408163265307e-06, | |
| "loss": 0.5738, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.559834361076355, | |
| "step": 60, | |
| "valid_targets_mean": 4199.9, | |
| "valid_targets_min": 1122 | |
| }, | |
| { | |
| "epoch": 0.11627906976744186, | |
| "grad_norm": 0.8020567416774592, | |
| "learning_rate": 6.530612244897959e-06, | |
| "loss": 0.5668, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.5514993071556091, | |
| "step": 65, | |
| "valid_targets_mean": 4092.1, | |
| "valid_targets_min": 1833 | |
| }, | |
| { | |
| "epoch": 0.1252236135957066, | |
| "grad_norm": 0.7566279895205129, | |
| "learning_rate": 7.0408163265306125e-06, | |
| "loss": 0.5578, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.5654456615447998, | |
| "step": 70, | |
| "valid_targets_mean": 4476.8, | |
| "valid_targets_min": 1837 | |
| }, | |
| { | |
| "epoch": 0.13416815742397137, | |
| "grad_norm": 0.8277372126838092, | |
| "learning_rate": 7.551020408163265e-06, | |
| "loss": 0.5356, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.4986897110939026, | |
| "step": 75, | |
| "valid_targets_mean": 3277.0, | |
| "valid_targets_min": 1319 | |
| }, | |
| { | |
| "epoch": 0.14311270125223613, | |
| "grad_norm": 0.7706280252042736, | |
| "learning_rate": 8.06122448979592e-06, | |
| "loss": 0.5068, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.5114747285842896, | |
| "step": 80, | |
| "valid_targets_mean": 3939.2, | |
| "valid_targets_min": 1961 | |
| }, | |
| { | |
| "epoch": 0.1520572450805009, | |
| "grad_norm": 0.7370945804564825, | |
| "learning_rate": 8.571428571428571e-06, | |
| "loss": 0.5039, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.521478533744812, | |
| "step": 85, | |
| "valid_targets_mean": 3797.7, | |
| "valid_targets_min": 864 | |
| }, | |
| { | |
| "epoch": 0.16100178890876565, | |
| "grad_norm": 0.6156667263403908, | |
| "learning_rate": 9.081632653061225e-06, | |
| "loss": 0.4835, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.4676780104637146, | |
| "step": 90, | |
| "valid_targets_mean": 4991.2, | |
| "valid_targets_min": 1062 | |
| }, | |
| { | |
| "epoch": 0.16994633273703041, | |
| "grad_norm": 0.6654121843319102, | |
| "learning_rate": 9.591836734693878e-06, | |
| "loss": 0.4785, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.48915815353393555, | |
| "step": 95, | |
| "valid_targets_mean": 3783.8, | |
| "valid_targets_min": 705 | |
| }, | |
| { | |
| "epoch": 0.17889087656529518, | |
| "grad_norm": 0.5656478848830337, | |
| "learning_rate": 1.0102040816326531e-05, | |
| "loss": 0.4789, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.4592038094997406, | |
| "step": 100, | |
| "valid_targets_mean": 4946.6, | |
| "valid_targets_min": 743 | |
| }, | |
| { | |
| "epoch": 0.18783542039355994, | |
| "grad_norm": 0.5753964943707206, | |
| "learning_rate": 1.0612244897959186e-05, | |
| "loss": 0.4481, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.4560643434524536, | |
| "step": 105, | |
| "valid_targets_mean": 4798.2, | |
| "valid_targets_min": 1049 | |
| }, | |
| { | |
| "epoch": 0.1967799642218247, | |
| "grad_norm": 0.6637720894363032, | |
| "learning_rate": 1.1122448979591838e-05, | |
| "loss": 0.4584, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.47521623969078064, | |
| "step": 110, | |
| "valid_targets_mean": 3852.6, | |
| "valid_targets_min": 1835 | |
| }, | |
| { | |
| "epoch": 0.20572450805008943, | |
| "grad_norm": 0.6937897615661168, | |
| "learning_rate": 1.1632653061224491e-05, | |
| "loss": 0.4686, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.45546483993530273, | |
| "step": 115, | |
| "valid_targets_mean": 4407.1, | |
| "valid_targets_min": 2317 | |
| }, | |
| { | |
| "epoch": 0.2146690518783542, | |
| "grad_norm": 0.6387384496375781, | |
| "learning_rate": 1.2142857142857142e-05, | |
| "loss": 0.4378, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.438065767288208, | |
| "step": 120, | |
| "valid_targets_mean": 4181.9, | |
| "valid_targets_min": 1324 | |
| }, | |
| { | |
| "epoch": 0.22361359570661896, | |
| "grad_norm": 0.6327727639217386, | |
| "learning_rate": 1.2653061224489798e-05, | |
| "loss": 0.4679, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.47834330797195435, | |
| "step": 125, | |
| "valid_targets_mean": 4268.0, | |
| "valid_targets_min": 1629 | |
| }, | |
| { | |
| "epoch": 0.23255813953488372, | |
| "grad_norm": 0.6622093710086933, | |
| "learning_rate": 1.316326530612245e-05, | |
| "loss": 0.427, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.4127560257911682, | |
| "step": 130, | |
| "valid_targets_mean": 3824.6, | |
| "valid_targets_min": 1371 | |
| }, | |
| { | |
| "epoch": 0.24150268336314848, | |
| "grad_norm": 0.6285981733100382, | |
| "learning_rate": 1.3673469387755102e-05, | |
| "loss": 0.4394, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.4288730323314667, | |
| "step": 135, | |
| "valid_targets_mean": 3877.5, | |
| "valid_targets_min": 1614 | |
| }, | |
| { | |
| "epoch": 0.2504472271914132, | |
| "grad_norm": 0.7388339921347419, | |
| "learning_rate": 1.4183673469387755e-05, | |
| "loss": 0.4429, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.420998752117157, | |
| "step": 140, | |
| "valid_targets_mean": 3424.4, | |
| "valid_targets_min": 780 | |
| }, | |
| { | |
| "epoch": 0.259391771019678, | |
| "grad_norm": 0.7101396979725441, | |
| "learning_rate": 1.469387755102041e-05, | |
| "loss": 0.4389, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.4431981146335602, | |
| "step": 145, | |
| "valid_targets_mean": 3347.7, | |
| "valid_targets_min": 1109 | |
| }, | |
| { | |
| "epoch": 0.26833631484794274, | |
| "grad_norm": 0.6957496254556903, | |
| "learning_rate": 1.5204081632653063e-05, | |
| "loss": 0.4362, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.4714646339416504, | |
| "step": 150, | |
| "valid_targets_mean": 4095.1, | |
| "valid_targets_min": 787 | |
| }, | |
| { | |
| "epoch": 0.2772808586762075, | |
| "grad_norm": 0.6624013827511412, | |
| "learning_rate": 1.5714285714285715e-05, | |
| "loss": 0.4158, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.4100586771965027, | |
| "step": 155, | |
| "valid_targets_mean": 3521.4, | |
| "valid_targets_min": 1542 | |
| }, | |
| { | |
| "epoch": 0.28622540250447226, | |
| "grad_norm": 1.0921780130652612, | |
| "learning_rate": 1.6224489795918368e-05, | |
| "loss": 0.4217, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.44413700699806213, | |
| "step": 160, | |
| "valid_targets_mean": 3212.3, | |
| "valid_targets_min": 945 | |
| }, | |
| { | |
| "epoch": 0.295169946332737, | |
| "grad_norm": 0.6555406982106572, | |
| "learning_rate": 1.673469387755102e-05, | |
| "loss": 0.411, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.43597084283828735, | |
| "step": 165, | |
| "valid_targets_mean": 4127.6, | |
| "valid_targets_min": 732 | |
| }, | |
| { | |
| "epoch": 0.3041144901610018, | |
| "grad_norm": 0.7156947764144866, | |
| "learning_rate": 1.7244897959183674e-05, | |
| "loss": 0.4142, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.45090579986572266, | |
| "step": 170, | |
| "valid_targets_mean": 4241.0, | |
| "valid_targets_min": 945 | |
| }, | |
| { | |
| "epoch": 0.31305903398926654, | |
| "grad_norm": 0.7843406742359783, | |
| "learning_rate": 1.7755102040816327e-05, | |
| "loss": 0.4088, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.409368097782135, | |
| "step": 175, | |
| "valid_targets_mean": 3534.1, | |
| "valid_targets_min": 607 | |
| }, | |
| { | |
| "epoch": 0.3220035778175313, | |
| "grad_norm": 0.6832085964985579, | |
| "learning_rate": 1.826530612244898e-05, | |
| "loss": 0.4323, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.4517691731452942, | |
| "step": 180, | |
| "valid_targets_mean": 4510.8, | |
| "valid_targets_min": 1674 | |
| }, | |
| { | |
| "epoch": 0.33094812164579607, | |
| "grad_norm": 0.6430408307738346, | |
| "learning_rate": 1.8775510204081636e-05, | |
| "loss": 0.406, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.4193289875984192, | |
| "step": 185, | |
| "valid_targets_mean": 4455.5, | |
| "valid_targets_min": 1476 | |
| }, | |
| { | |
| "epoch": 0.33989266547406083, | |
| "grad_norm": 0.6803856005045361, | |
| "learning_rate": 1.928571428571429e-05, | |
| "loss": 0.3807, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.376784086227417, | |
| "step": 190, | |
| "valid_targets_mean": 3769.6, | |
| "valid_targets_min": 715 | |
| }, | |
| { | |
| "epoch": 0.3488372093023256, | |
| "grad_norm": 0.7962416214555187, | |
| "learning_rate": 1.979591836734694e-05, | |
| "loss": 0.4095, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.40137556195259094, | |
| "step": 195, | |
| "valid_targets_mean": 3348.6, | |
| "valid_targets_min": 831 | |
| }, | |
| { | |
| "epoch": 0.35778175313059035, | |
| "grad_norm": 0.6163018468674557, | |
| "learning_rate": 2.0306122448979594e-05, | |
| "loss": 0.4097, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.4052973985671997, | |
| "step": 200, | |
| "valid_targets_mean": 4064.9, | |
| "valid_targets_min": 928 | |
| }, | |
| { | |
| "epoch": 0.3667262969588551, | |
| "grad_norm": 0.5825225301762951, | |
| "learning_rate": 2.0816326530612247e-05, | |
| "loss": 0.3895, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.379678338766098, | |
| "step": 205, | |
| "valid_targets_mean": 4561.4, | |
| "valid_targets_min": 1682 | |
| }, | |
| { | |
| "epoch": 0.3756708407871199, | |
| "grad_norm": 0.7186506278507635, | |
| "learning_rate": 2.13265306122449e-05, | |
| "loss": 0.3987, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3955579996109009, | |
| "step": 210, | |
| "valid_targets_mean": 3334.4, | |
| "valid_targets_min": 877 | |
| }, | |
| { | |
| "epoch": 0.38461538461538464, | |
| "grad_norm": 0.7102346399517735, | |
| "learning_rate": 2.1836734693877552e-05, | |
| "loss": 0.3996, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3910446763038635, | |
| "step": 215, | |
| "valid_targets_mean": 3237.5, | |
| "valid_targets_min": 1604 | |
| }, | |
| { | |
| "epoch": 0.3935599284436494, | |
| "grad_norm": 0.6261071087270799, | |
| "learning_rate": 2.2346938775510205e-05, | |
| "loss": 0.3979, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.371921181678772, | |
| "step": 220, | |
| "valid_targets_mean": 4290.6, | |
| "valid_targets_min": 1637 | |
| }, | |
| { | |
| "epoch": 0.40250447227191416, | |
| "grad_norm": 0.6481717381436756, | |
| "learning_rate": 2.2857142857142858e-05, | |
| "loss": 0.3915, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3760562539100647, | |
| "step": 225, | |
| "valid_targets_mean": 3945.3, | |
| "valid_targets_min": 1767 | |
| }, | |
| { | |
| "epoch": 0.41144901610017887, | |
| "grad_norm": 0.6201712385655748, | |
| "learning_rate": 2.336734693877551e-05, | |
| "loss": 0.3929, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.4008695185184479, | |
| "step": 230, | |
| "valid_targets_mean": 4555.8, | |
| "valid_targets_min": 1312 | |
| }, | |
| { | |
| "epoch": 0.4203935599284436, | |
| "grad_norm": 0.6203763910651068, | |
| "learning_rate": 2.3877551020408167e-05, | |
| "loss": 0.3889, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.41413581371307373, | |
| "step": 235, | |
| "valid_targets_mean": 4777.6, | |
| "valid_targets_min": 2200 | |
| }, | |
| { | |
| "epoch": 0.4293381037567084, | |
| "grad_norm": 0.65351087223048, | |
| "learning_rate": 2.438775510204082e-05, | |
| "loss": 0.3809, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3872208297252655, | |
| "step": 240, | |
| "valid_targets_mean": 3722.7, | |
| "valid_targets_min": 1608 | |
| }, | |
| { | |
| "epoch": 0.43828264758497315, | |
| "grad_norm": 0.6802074572655615, | |
| "learning_rate": 2.4897959183673473e-05, | |
| "loss": 0.3602, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.36208364367485046, | |
| "step": 245, | |
| "valid_targets_mean": 4687.7, | |
| "valid_targets_min": 1944 | |
| }, | |
| { | |
| "epoch": 0.4472271914132379, | |
| "grad_norm": 0.8869715993980485, | |
| "learning_rate": 2.5408163265306125e-05, | |
| "loss": 0.3921, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.4071674942970276, | |
| "step": 250, | |
| "valid_targets_mean": 4066.7, | |
| "valid_targets_min": 1633 | |
| }, | |
| { | |
| "epoch": 0.4561717352415027, | |
| "grad_norm": 0.7031570375051375, | |
| "learning_rate": 2.5918367346938778e-05, | |
| "loss": 0.3945, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.38449206948280334, | |
| "step": 255, | |
| "valid_targets_mean": 3275.8, | |
| "valid_targets_min": 807 | |
| }, | |
| { | |
| "epoch": 0.46511627906976744, | |
| "grad_norm": 0.6401965700906631, | |
| "learning_rate": 2.642857142857143e-05, | |
| "loss": 0.3799, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.4059576094150543, | |
| "step": 260, | |
| "valid_targets_mean": 4899.2, | |
| "valid_targets_min": 1869 | |
| }, | |
| { | |
| "epoch": 0.4740608228980322, | |
| "grad_norm": 0.68848774995671, | |
| "learning_rate": 2.6938775510204084e-05, | |
| "loss": 0.3899, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.42488914728164673, | |
| "step": 265, | |
| "valid_targets_mean": 4040.0, | |
| "valid_targets_min": 1784 | |
| }, | |
| { | |
| "epoch": 0.48300536672629696, | |
| "grad_norm": 0.6555186879962056, | |
| "learning_rate": 2.7448979591836737e-05, | |
| "loss": 0.3747, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.37860578298568726, | |
| "step": 270, | |
| "valid_targets_mean": 3953.7, | |
| "valid_targets_min": 1120 | |
| }, | |
| { | |
| "epoch": 0.4919499105545617, | |
| "grad_norm": 0.6958637124427978, | |
| "learning_rate": 2.7959183673469393e-05, | |
| "loss": 0.3657, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.353659987449646, | |
| "step": 275, | |
| "valid_targets_mean": 4198.6, | |
| "valid_targets_min": 867 | |
| }, | |
| { | |
| "epoch": 0.5008944543828264, | |
| "grad_norm": 0.7270956530331304, | |
| "learning_rate": 2.8469387755102046e-05, | |
| "loss": 0.3695, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3616817593574524, | |
| "step": 280, | |
| "valid_targets_mean": 3319.2, | |
| "valid_targets_min": 1008 | |
| }, | |
| { | |
| "epoch": 0.5098389982110912, | |
| "grad_norm": 0.6584523988533558, | |
| "learning_rate": 2.89795918367347e-05, | |
| "loss": 0.3639, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3544594645500183, | |
| "step": 285, | |
| "valid_targets_mean": 3901.2, | |
| "valid_targets_min": 1630 | |
| }, | |
| { | |
| "epoch": 0.518783542039356, | |
| "grad_norm": 0.6603276785502654, | |
| "learning_rate": 2.948979591836735e-05, | |
| "loss": 0.3765, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3783223032951355, | |
| "step": 290, | |
| "valid_targets_mean": 4002.2, | |
| "valid_targets_min": 1031 | |
| }, | |
| { | |
| "epoch": 0.5277280858676208, | |
| "grad_norm": 0.6256302987866349, | |
| "learning_rate": 3.0000000000000004e-05, | |
| "loss": 0.3801, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3801421523094177, | |
| "step": 295, | |
| "valid_targets_mean": 4247.8, | |
| "valid_targets_min": 825 | |
| }, | |
| { | |
| "epoch": 0.5366726296958855, | |
| "grad_norm": 0.641725168293166, | |
| "learning_rate": 3.0510204081632654e-05, | |
| "loss": 0.3661, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3630704879760742, | |
| "step": 300, | |
| "valid_targets_mean": 3875.1, | |
| "valid_targets_min": 1685 | |
| }, | |
| { | |
| "epoch": 0.5456171735241503, | |
| "grad_norm": 0.811210240684181, | |
| "learning_rate": 3.102040816326531e-05, | |
| "loss": 0.3741, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.36272579431533813, | |
| "step": 305, | |
| "valid_targets_mean": 3257.9, | |
| "valid_targets_min": 1264 | |
| }, | |
| { | |
| "epoch": 0.554561717352415, | |
| "grad_norm": 0.667191922839525, | |
| "learning_rate": 3.153061224489796e-05, | |
| "loss": 0.3667, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3673118054866791, | |
| "step": 310, | |
| "valid_targets_mean": 4112.9, | |
| "valid_targets_min": 923 | |
| }, | |
| { | |
| "epoch": 0.5635062611806798, | |
| "grad_norm": 0.6483488033218554, | |
| "learning_rate": 3.2040816326530615e-05, | |
| "loss": 0.3695, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3770442008972168, | |
| "step": 315, | |
| "valid_targets_mean": 4392.9, | |
| "valid_targets_min": 2212 | |
| }, | |
| { | |
| "epoch": 0.5724508050089445, | |
| "grad_norm": 0.6594195347781112, | |
| "learning_rate": 3.255102040816327e-05, | |
| "loss": 0.3851, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3894268274307251, | |
| "step": 320, | |
| "valid_targets_mean": 4147.8, | |
| "valid_targets_min": 1795 | |
| }, | |
| { | |
| "epoch": 0.5813953488372093, | |
| "grad_norm": 0.5800198159071142, | |
| "learning_rate": 3.306122448979592e-05, | |
| "loss": 0.3612, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3446458876132965, | |
| "step": 325, | |
| "valid_targets_mean": 4526.3, | |
| "valid_targets_min": 1608 | |
| }, | |
| { | |
| "epoch": 0.590339892665474, | |
| "grad_norm": 0.6476011130202899, | |
| "learning_rate": 3.357142857142858e-05, | |
| "loss": 0.3631, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.36270543932914734, | |
| "step": 330, | |
| "valid_targets_mean": 4024.4, | |
| "valid_targets_min": 1450 | |
| }, | |
| { | |
| "epoch": 0.5992844364937389, | |
| "grad_norm": 0.6522027257372564, | |
| "learning_rate": 3.408163265306123e-05, | |
| "loss": 0.3654, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3855012357234955, | |
| "step": 335, | |
| "valid_targets_mean": 4163.4, | |
| "valid_targets_min": 1675 | |
| }, | |
| { | |
| "epoch": 0.6082289803220036, | |
| "grad_norm": 0.6243810739429985, | |
| "learning_rate": 3.459183673469388e-05, | |
| "loss": 0.3671, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3456933796405792, | |
| "step": 340, | |
| "valid_targets_mean": 3930.6, | |
| "valid_targets_min": 1588 | |
| }, | |
| { | |
| "epoch": 0.6171735241502684, | |
| "grad_norm": 0.6158815913365016, | |
| "learning_rate": 3.510204081632653e-05, | |
| "loss": 0.3655, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.35427260398864746, | |
| "step": 345, | |
| "valid_targets_mean": 4795.0, | |
| "valid_targets_min": 1934 | |
| }, | |
| { | |
| "epoch": 0.6261180679785331, | |
| "grad_norm": 0.7068523423045154, | |
| "learning_rate": 3.561224489795918e-05, | |
| "loss": 0.3687, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.4106198847293854, | |
| "step": 350, | |
| "valid_targets_mean": 4016.4, | |
| "valid_targets_min": 648 | |
| }, | |
| { | |
| "epoch": 0.6350626118067979, | |
| "grad_norm": 0.7843744948487554, | |
| "learning_rate": 3.612244897959184e-05, | |
| "loss": 0.3634, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.36276382207870483, | |
| "step": 355, | |
| "valid_targets_mean": 3936.0, | |
| "valid_targets_min": 1615 | |
| }, | |
| { | |
| "epoch": 0.6440071556350626, | |
| "grad_norm": 0.6872061405347063, | |
| "learning_rate": 3.6632653061224494e-05, | |
| "loss": 0.3486, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.33801645040512085, | |
| "step": 360, | |
| "valid_targets_mean": 3690.8, | |
| "valid_targets_min": 1805 | |
| }, | |
| { | |
| "epoch": 0.6529516994633273, | |
| "grad_norm": 0.626374760633224, | |
| "learning_rate": 3.714285714285715e-05, | |
| "loss": 0.3685, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3595544695854187, | |
| "step": 365, | |
| "valid_targets_mean": 4637.8, | |
| "valid_targets_min": 621 | |
| }, | |
| { | |
| "epoch": 0.6618962432915921, | |
| "grad_norm": 0.6319521644936286, | |
| "learning_rate": 3.76530612244898e-05, | |
| "loss": 0.3652, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.37265145778656006, | |
| "step": 370, | |
| "valid_targets_mean": 4238.1, | |
| "valid_targets_min": 1087 | |
| }, | |
| { | |
| "epoch": 0.6708407871198568, | |
| "grad_norm": 0.6238512565906754, | |
| "learning_rate": 3.816326530612245e-05, | |
| "loss": 0.3527, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.36174702644348145, | |
| "step": 375, | |
| "valid_targets_mean": 4511.5, | |
| "valid_targets_min": 1518 | |
| }, | |
| { | |
| "epoch": 0.6797853309481217, | |
| "grad_norm": 0.569965173329661, | |
| "learning_rate": 3.8673469387755105e-05, | |
| "loss": 0.3637, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.34279459714889526, | |
| "step": 380, | |
| "valid_targets_mean": 4155.4, | |
| "valid_targets_min": 1851 | |
| }, | |
| { | |
| "epoch": 0.6887298747763864, | |
| "grad_norm": 0.6707477926009685, | |
| "learning_rate": 3.9183673469387755e-05, | |
| "loss": 0.3725, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3640945553779602, | |
| "step": 385, | |
| "valid_targets_mean": 4214.8, | |
| "valid_targets_min": 767 | |
| }, | |
| { | |
| "epoch": 0.6976744186046512, | |
| "grad_norm": 0.5367271020119635, | |
| "learning_rate": 3.969387755102041e-05, | |
| "loss": 0.3422, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3215610384941101, | |
| "step": 390, | |
| "valid_targets_mean": 4804.4, | |
| "valid_targets_min": 1580 | |
| }, | |
| { | |
| "epoch": 0.7066189624329159, | |
| "grad_norm": 0.6919691468354696, | |
| "learning_rate": 3.9999968156003224e-05, | |
| "loss": 0.3659, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3643749952316284, | |
| "step": 395, | |
| "valid_targets_mean": 3619.4, | |
| "valid_targets_min": 993 | |
| }, | |
| { | |
| "epoch": 0.7155635062611807, | |
| "grad_norm": 0.6288859512995435, | |
| "learning_rate": 3.999960991220401e-05, | |
| "loss": 0.3524, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3619106709957123, | |
| "step": 400, | |
| "valid_targets_mean": 4227.4, | |
| "valid_targets_min": 2088 | |
| }, | |
| { | |
| "epoch": 0.7245080500894454, | |
| "grad_norm": 0.6317063641588481, | |
| "learning_rate": 3.9998853626763316e-05, | |
| "loss": 0.3548, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3588660657405853, | |
| "step": 405, | |
| "valid_targets_mean": 4536.9, | |
| "valid_targets_min": 1735 | |
| }, | |
| { | |
| "epoch": 0.7334525939177102, | |
| "grad_norm": 0.6057461041737375, | |
| "learning_rate": 3.999769931473309e-05, | |
| "loss": 0.3654, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3536277711391449, | |
| "step": 410, | |
| "valid_targets_mean": 3928.1, | |
| "valid_targets_min": 1408 | |
| }, | |
| { | |
| "epoch": 0.7423971377459749, | |
| "grad_norm": 0.5865439635131978, | |
| "learning_rate": 3.999614699908698e-05, | |
| "loss": 0.3447, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3287227153778076, | |
| "step": 415, | |
| "valid_targets_mean": 4680.5, | |
| "valid_targets_min": 1738 | |
| }, | |
| { | |
| "epoch": 0.7513416815742398, | |
| "grad_norm": 0.6236369468759124, | |
| "learning_rate": 3.999419671071993e-05, | |
| "loss": 0.36, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3473414182662964, | |
| "step": 420, | |
| "valid_targets_mean": 4636.2, | |
| "valid_targets_min": 1483 | |
| }, | |
| { | |
| "epoch": 0.7602862254025045, | |
| "grad_norm": 0.6239705685634717, | |
| "learning_rate": 3.999184848844746e-05, | |
| "loss": 0.3558, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.31550854444503784, | |
| "step": 425, | |
| "valid_targets_mean": 3627.9, | |
| "valid_targets_min": 1473 | |
| }, | |
| { | |
| "epoch": 0.7692307692307693, | |
| "grad_norm": 0.7215363445685998, | |
| "learning_rate": 3.9989102379005026e-05, | |
| "loss": 0.3531, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3723289370536804, | |
| "step": 430, | |
| "valid_targets_mean": 3313.8, | |
| "valid_targets_min": 777 | |
| }, | |
| { | |
| "epoch": 0.778175313059034, | |
| "grad_norm": 0.6314029660945639, | |
| "learning_rate": 3.9985958437046976e-05, | |
| "loss": 0.346, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3319028913974762, | |
| "step": 435, | |
| "valid_targets_mean": 4074.1, | |
| "valid_targets_min": 1150 | |
| }, | |
| { | |
| "epoch": 0.7871198568872988, | |
| "grad_norm": 0.671810871870371, | |
| "learning_rate": 3.998241672514551e-05, | |
| "loss": 0.3593, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3606319725513458, | |
| "step": 440, | |
| "valid_targets_mean": 3716.7, | |
| "valid_targets_min": 1462 | |
| }, | |
| { | |
| "epoch": 0.7960644007155635, | |
| "grad_norm": 0.6534639604048026, | |
| "learning_rate": 3.997847731378946e-05, | |
| "loss": 0.3545, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3522427976131439, | |
| "step": 445, | |
| "valid_targets_mean": 3853.1, | |
| "valid_targets_min": 1761 | |
| }, | |
| { | |
| "epoch": 0.8050089445438283, | |
| "grad_norm": 0.6256271897448633, | |
| "learning_rate": 3.9974140281382844e-05, | |
| "loss": 0.3523, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3305948078632355, | |
| "step": 450, | |
| "valid_targets_mean": 4348.7, | |
| "valid_targets_min": 2006 | |
| }, | |
| { | |
| "epoch": 0.813953488372093, | |
| "grad_norm": 0.665427663210199, | |
| "learning_rate": 3.996940571424331e-05, | |
| "loss": 0.3417, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3263488709926605, | |
| "step": 455, | |
| "valid_targets_mean": 3719.2, | |
| "valid_targets_min": 2056 | |
| }, | |
| { | |
| "epoch": 0.8228980322003577, | |
| "grad_norm": 0.6525668365849738, | |
| "learning_rate": 3.996427370660045e-05, | |
| "loss": 0.3491, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.357142835855484, | |
| "step": 460, | |
| "valid_targets_mean": 4044.6, | |
| "valid_targets_min": 681 | |
| }, | |
| { | |
| "epoch": 0.8318425760286225, | |
| "grad_norm": 0.5874893016038123, | |
| "learning_rate": 3.995874436059389e-05, | |
| "loss": 0.3463, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.34184402227401733, | |
| "step": 465, | |
| "valid_targets_mean": 4344.1, | |
| "valid_targets_min": 1744 | |
| }, | |
| { | |
| "epoch": 0.8407871198568873, | |
| "grad_norm": 0.6543028013713276, | |
| "learning_rate": 3.9952817786271264e-05, | |
| "loss": 0.3456, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3407524824142456, | |
| "step": 470, | |
| "valid_targets_mean": 4510.4, | |
| "valid_targets_min": 2428 | |
| }, | |
| { | |
| "epoch": 0.8497316636851521, | |
| "grad_norm": 0.6482522573193757, | |
| "learning_rate": 3.994649410158605e-05, | |
| "loss": 0.3445, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.336134672164917, | |
| "step": 475, | |
| "valid_targets_mean": 3787.4, | |
| "valid_targets_min": 545 | |
| }, | |
| { | |
| "epoch": 0.8586762075134168, | |
| "grad_norm": 0.6618318888035204, | |
| "learning_rate": 3.99397734323952e-05, | |
| "loss": 0.3559, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.39081141352653503, | |
| "step": 480, | |
| "valid_targets_mean": 4617.5, | |
| "valid_targets_min": 778 | |
| }, | |
| { | |
| "epoch": 0.8676207513416816, | |
| "grad_norm": 0.6461883560383084, | |
| "learning_rate": 3.993265591245662e-05, | |
| "loss": 0.3481, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3543085753917694, | |
| "step": 485, | |
| "valid_targets_mean": 4650.8, | |
| "valid_targets_min": 1304 | |
| }, | |
| { | |
| "epoch": 0.8765652951699463, | |
| "grad_norm": 0.7166558381651617, | |
| "learning_rate": 3.992514168342655e-05, | |
| "loss": 0.355, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3686150312423706, | |
| "step": 490, | |
| "valid_targets_mean": 3723.8, | |
| "valid_targets_min": 927 | |
| }, | |
| { | |
| "epoch": 0.8855098389982111, | |
| "grad_norm": 0.6815294274219248, | |
| "learning_rate": 3.9917230894856705e-05, | |
| "loss": 0.3437, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3497094511985779, | |
| "step": 495, | |
| "valid_targets_mean": 3219.9, | |
| "valid_targets_min": 669 | |
| }, | |
| { | |
| "epoch": 0.8944543828264758, | |
| "grad_norm": 0.78386232455594, | |
| "learning_rate": 3.990892370419132e-05, | |
| "loss": 0.3493, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.35013243556022644, | |
| "step": 500, | |
| "valid_targets_mean": 3843.9, | |
| "valid_targets_min": 1370 | |
| }, | |
| { | |
| "epoch": 0.9033989266547406, | |
| "grad_norm": 0.6199422182618969, | |
| "learning_rate": 3.9900220276764013e-05, | |
| "loss": 0.3418, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.35459715127944946, | |
| "step": 505, | |
| "valid_targets_mean": 4905.9, | |
| "valid_targets_min": 2222 | |
| }, | |
| { | |
| "epoch": 0.9123434704830053, | |
| "grad_norm": 0.6761601566190695, | |
| "learning_rate": 3.989112078579449e-05, | |
| "loss": 0.3363, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.34386712312698364, | |
| "step": 510, | |
| "valid_targets_mean": 3200.0, | |
| "valid_targets_min": 1484 | |
| }, | |
| { | |
| "epoch": 0.9212880143112702, | |
| "grad_norm": 0.5809388892974982, | |
| "learning_rate": 3.988162541238509e-05, | |
| "loss": 0.3556, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3644087314605713, | |
| "step": 515, | |
| "valid_targets_mean": 4901.0, | |
| "valid_targets_min": 720 | |
| }, | |
| { | |
| "epoch": 0.9302325581395349, | |
| "grad_norm": 0.6670957667841473, | |
| "learning_rate": 3.98717343455172e-05, | |
| "loss": 0.3468, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3259006440639496, | |
| "step": 520, | |
| "valid_targets_mean": 3929.0, | |
| "valid_targets_min": 819 | |
| }, | |
| { | |
| "epoch": 0.9391771019677997, | |
| "grad_norm": 0.6133306725020052, | |
| "learning_rate": 3.9861447782047495e-05, | |
| "loss": 0.3521, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3232124149799347, | |
| "step": 525, | |
| "valid_targets_mean": 4338.9, | |
| "valid_targets_min": 1621 | |
| }, | |
| { | |
| "epoch": 0.9481216457960644, | |
| "grad_norm": 0.5945215828915712, | |
| "learning_rate": 3.9850765926704e-05, | |
| "loss": 0.3457, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3464571535587311, | |
| "step": 530, | |
| "valid_targets_mean": 4751.4, | |
| "valid_targets_min": 1880 | |
| }, | |
| { | |
| "epoch": 0.9570661896243292, | |
| "grad_norm": 0.616855899500005, | |
| "learning_rate": 3.9839688992082004e-05, | |
| "loss": 0.3406, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.33732545375823975, | |
| "step": 535, | |
| "valid_targets_mean": 3987.8, | |
| "valid_targets_min": 1435 | |
| }, | |
| { | |
| "epoch": 0.9660107334525939, | |
| "grad_norm": 0.5714623187800117, | |
| "learning_rate": 3.9828217198639884e-05, | |
| "loss": 0.345, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3435646891593933, | |
| "step": 540, | |
| "valid_targets_mean": 4915.8, | |
| "valid_targets_min": 1400 | |
| }, | |
| { | |
| "epoch": 0.9749552772808586, | |
| "grad_norm": 0.6324746280014568, | |
| "learning_rate": 3.981635077469468e-05, | |
| "loss": 0.3394, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3339131474494934, | |
| "step": 545, | |
| "valid_targets_mean": 3912.2, | |
| "valid_targets_min": 1666 | |
| }, | |
| { | |
| "epoch": 0.9838998211091234, | |
| "grad_norm": 0.626118262073981, | |
| "learning_rate": 3.980408995641751e-05, | |
| "loss": 0.3219, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.33410966396331787, | |
| "step": 550, | |
| "valid_targets_mean": 4326.9, | |
| "valid_targets_min": 1565 | |
| }, | |
| { | |
| "epoch": 0.9928443649373881, | |
| "grad_norm": 0.5559330654712583, | |
| "learning_rate": 3.979143498782898e-05, | |
| "loss": 0.3523, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3249479830265045, | |
| "step": 555, | |
| "valid_targets_mean": 4442.0, | |
| "valid_targets_min": 1579 | |
| }, | |
| { | |
| "epoch": 1.0017889087656529, | |
| "grad_norm": 0.5744899255811827, | |
| "learning_rate": 3.977838612079419e-05, | |
| "loss": 0.3482, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.34298670291900635, | |
| "step": 560, | |
| "valid_targets_mean": 4366.9, | |
| "valid_targets_min": 1630 | |
| }, | |
| { | |
| "epoch": 1.0107334525939178, | |
| "grad_norm": 0.540064997986571, | |
| "learning_rate": 3.976494361501786e-05, | |
| "loss": 0.3255, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.29019173979759216, | |
| "step": 565, | |
| "valid_targets_mean": 4183.6, | |
| "valid_targets_min": 1293 | |
| }, | |
| { | |
| "epoch": 1.0196779964221825, | |
| "grad_norm": 0.6708126937194594, | |
| "learning_rate": 3.975110773803904e-05, | |
| "loss": 0.3291, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3077584505081177, | |
| "step": 570, | |
| "valid_targets_mean": 3365.9, | |
| "valid_targets_min": 1599 | |
| }, | |
| { | |
| "epoch": 1.0286225402504472, | |
| "grad_norm": 0.5975546705521818, | |
| "learning_rate": 3.973687876522587e-05, | |
| "loss": 0.3099, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3131313920021057, | |
| "step": 575, | |
| "valid_targets_mean": 3975.9, | |
| "valid_targets_min": 1659 | |
| }, | |
| { | |
| "epoch": 1.037567084078712, | |
| "grad_norm": 0.5802313523133982, | |
| "learning_rate": 3.9722256979770054e-05, | |
| "loss": 0.3228, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3223347067832947, | |
| "step": 580, | |
| "valid_targets_mean": 3876.8, | |
| "valid_targets_min": 1256 | |
| }, | |
| { | |
| "epoch": 1.0465116279069768, | |
| "grad_norm": 0.5940502139996998, | |
| "learning_rate": 3.970724267268125e-05, | |
| "loss": 0.3316, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3173867166042328, | |
| "step": 585, | |
| "valid_targets_mean": 3916.0, | |
| "valid_targets_min": 1519 | |
| }, | |
| { | |
| "epoch": 1.0554561717352415, | |
| "grad_norm": 0.5526882078589692, | |
| "learning_rate": 3.969183614278125e-05, | |
| "loss": 0.3205, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3090658187866211, | |
| "step": 590, | |
| "valid_targets_mean": 3783.8, | |
| "valid_targets_min": 1558 | |
| }, | |
| { | |
| "epoch": 1.0644007155635062, | |
| "grad_norm": 0.5756072638500698, | |
| "learning_rate": 3.9676037696698056e-05, | |
| "loss": 0.3451, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.33000075817108154, | |
| "step": 595, | |
| "valid_targets_mean": 4377.0, | |
| "valid_targets_min": 1580 | |
| }, | |
| { | |
| "epoch": 1.073345259391771, | |
| "grad_norm": 0.6209850679632152, | |
| "learning_rate": 3.9659847648859775e-05, | |
| "loss": 0.3358, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3666132092475891, | |
| "step": 600, | |
| "valid_targets_mean": 3891.3, | |
| "valid_targets_min": 766 | |
| }, | |
| { | |
| "epoch": 1.0822898032200359, | |
| "grad_norm": 0.574730620347152, | |
| "learning_rate": 3.9643266321488354e-05, | |
| "loss": 0.3297, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.31989532709121704, | |
| "step": 605, | |
| "valid_targets_mean": 4488.6, | |
| "valid_targets_min": 1565 | |
| }, | |
| { | |
| "epoch": 1.0912343470483006, | |
| "grad_norm": 0.7137118300894771, | |
| "learning_rate": 3.962629404459317e-05, | |
| "loss": 0.3272, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3319159746170044, | |
| "step": 610, | |
| "valid_targets_mean": 2972.9, | |
| "valid_targets_min": 625 | |
| }, | |
| { | |
| "epoch": 1.1001788908765653, | |
| "grad_norm": 0.5664071308134078, | |
| "learning_rate": 3.960893115596445e-05, | |
| "loss": 0.3098, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2983168065547943, | |
| "step": 615, | |
| "valid_targets_mean": 4385.4, | |
| "valid_targets_min": 1484 | |
| }, | |
| { | |
| "epoch": 1.10912343470483, | |
| "grad_norm": 0.617788180227712, | |
| "learning_rate": 3.959117800116658e-05, | |
| "loss": 0.3321, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3224518299102783, | |
| "step": 620, | |
| "valid_targets_mean": 3728.5, | |
| "valid_targets_min": 988 | |
| }, | |
| { | |
| "epoch": 1.118067978533095, | |
| "grad_norm": 0.6068172667148498, | |
| "learning_rate": 3.9573034933531195e-05, | |
| "loss": 0.3308, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3229108154773712, | |
| "step": 625, | |
| "valid_targets_mean": 4483.8, | |
| "valid_targets_min": 1824 | |
| }, | |
| { | |
| "epoch": 1.1270125223613596, | |
| "grad_norm": 0.6113158435705944, | |
| "learning_rate": 3.955450231415014e-05, | |
| "loss": 0.3405, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3196060061454773, | |
| "step": 630, | |
| "valid_targets_mean": 4137.5, | |
| "valid_targets_min": 1364 | |
| }, | |
| { | |
| "epoch": 1.1359570661896243, | |
| "grad_norm": 0.6173710617052496, | |
| "learning_rate": 3.953558051186834e-05, | |
| "loss": 0.3333, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.34940510988235474, | |
| "step": 635, | |
| "valid_targets_mean": 4084.4, | |
| "valid_targets_min": 1886 | |
| }, | |
| { | |
| "epoch": 1.144901610017889, | |
| "grad_norm": 0.6846373488774041, | |
| "learning_rate": 3.95162699032764e-05, | |
| "loss": 0.3244, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3563871383666992, | |
| "step": 640, | |
| "valid_targets_mean": 4735.4, | |
| "valid_targets_min": 1188 | |
| }, | |
| { | |
| "epoch": 1.1538461538461537, | |
| "grad_norm": 0.6569856707625712, | |
| "learning_rate": 3.949657087270313e-05, | |
| "loss": 0.3199, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.32631808519363403, | |
| "step": 645, | |
| "valid_targets_mean": 4182.3, | |
| "valid_targets_min": 850 | |
| }, | |
| { | |
| "epoch": 1.1627906976744187, | |
| "grad_norm": 0.5672879733062905, | |
| "learning_rate": 3.947648381220789e-05, | |
| "loss": 0.3152, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.31764596700668335, | |
| "step": 650, | |
| "valid_targets_mean": 3908.7, | |
| "valid_targets_min": 646 | |
| }, | |
| { | |
| "epoch": 1.1717352415026834, | |
| "grad_norm": 0.6801671519180097, | |
| "learning_rate": 3.9456009121572824e-05, | |
| "loss": 0.3233, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.31651636958122253, | |
| "step": 655, | |
| "valid_targets_mean": 3767.4, | |
| "valid_targets_min": 1485 | |
| }, | |
| { | |
| "epoch": 1.180679785330948, | |
| "grad_norm": 0.5776768005096825, | |
| "learning_rate": 3.943514720829485e-05, | |
| "loss": 0.3302, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3585665225982666, | |
| "step": 660, | |
| "valid_targets_mean": 4321.3, | |
| "valid_targets_min": 1595 | |
| }, | |
| { | |
| "epoch": 1.1896243291592128, | |
| "grad_norm": 0.6632996529951345, | |
| "learning_rate": 3.941389848757756e-05, | |
| "loss": 0.3293, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.34740591049194336, | |
| "step": 665, | |
| "valid_targets_mean": 3854.9, | |
| "valid_targets_min": 1643 | |
| }, | |
| { | |
| "epoch": 1.1985688729874777, | |
| "grad_norm": 0.6568776669609784, | |
| "learning_rate": 3.9392263382323e-05, | |
| "loss": 0.3271, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3057899475097656, | |
| "step": 670, | |
| "valid_targets_mean": 3277.4, | |
| "valid_targets_min": 1615 | |
| }, | |
| { | |
| "epoch": 1.2075134168157424, | |
| "grad_norm": 0.6270834310864767, | |
| "learning_rate": 3.93702423231232e-05, | |
| "loss": 0.3227, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.32783782482147217, | |
| "step": 675, | |
| "valid_targets_mean": 3216.5, | |
| "valid_targets_min": 1055 | |
| }, | |
| { | |
| "epoch": 1.2164579606440071, | |
| "grad_norm": 0.6305095200057445, | |
| "learning_rate": 3.9347835748251645e-05, | |
| "loss": 0.332, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3536560535430908, | |
| "step": 680, | |
| "valid_targets_mean": 3572.4, | |
| "valid_targets_min": 552 | |
| }, | |
| { | |
| "epoch": 1.2254025044722718, | |
| "grad_norm": 0.5736045805917658, | |
| "learning_rate": 3.9325044103654526e-05, | |
| "loss": 0.3348, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3276847004890442, | |
| "step": 685, | |
| "valid_targets_mean": 3811.6, | |
| "valid_targets_min": 1374 | |
| }, | |
| { | |
| "epoch": 1.2343470483005368, | |
| "grad_norm": 0.5244473851418727, | |
| "learning_rate": 3.9301867842941867e-05, | |
| "loss": 0.3302, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3139815926551819, | |
| "step": 690, | |
| "valid_targets_mean": 4468.8, | |
| "valid_targets_min": 1504 | |
| }, | |
| { | |
| "epoch": 1.2432915921288015, | |
| "grad_norm": 0.5751467981153631, | |
| "learning_rate": 3.9278307427378495e-05, | |
| "loss": 0.335, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.320250928401947, | |
| "step": 695, | |
| "valid_targets_mean": 4114.9, | |
| "valid_targets_min": 2051 | |
| }, | |
| { | |
| "epoch": 1.2522361359570662, | |
| "grad_norm": 0.57724364766724, | |
| "learning_rate": 3.92543633258749e-05, | |
| "loss": 0.3328, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3237752914428711, | |
| "step": 700, | |
| "valid_targets_mean": 4272.7, | |
| "valid_targets_min": 1724 | |
| }, | |
| { | |
| "epoch": 1.2611806797853309, | |
| "grad_norm": 0.6222962371838422, | |
| "learning_rate": 3.923003601497785e-05, | |
| "loss": 0.3096, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2992010712623596, | |
| "step": 705, | |
| "valid_targets_mean": 3838.6, | |
| "valid_targets_min": 1073 | |
| }, | |
| { | |
| "epoch": 1.2701252236135958, | |
| "grad_norm": 0.5745002540877785, | |
| "learning_rate": 3.920532597886091e-05, | |
| "loss": 0.3273, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.33431553840637207, | |
| "step": 710, | |
| "valid_targets_mean": 4709.8, | |
| "valid_targets_min": 1660 | |
| }, | |
| { | |
| "epoch": 1.2790697674418605, | |
| "grad_norm": 0.8250895929655195, | |
| "learning_rate": 3.918023370931485e-05, | |
| "loss": 0.3325, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.34380948543548584, | |
| "step": 715, | |
| "valid_targets_mean": 3260.9, | |
| "valid_targets_min": 1452 | |
| }, | |
| { | |
| "epoch": 1.2880143112701252, | |
| "grad_norm": 0.5891395257155815, | |
| "learning_rate": 3.915475970573782e-05, | |
| "loss": 0.3116, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.31047481298446655, | |
| "step": 720, | |
| "valid_targets_mean": 3394.5, | |
| "valid_targets_min": 728 | |
| }, | |
| { | |
| "epoch": 1.29695885509839, | |
| "grad_norm": 0.5492069212467995, | |
| "learning_rate": 3.9128904475125414e-05, | |
| "loss": 0.3236, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3350293040275574, | |
| "step": 725, | |
| "valid_targets_mean": 4490.6, | |
| "valid_targets_min": 1470 | |
| }, | |
| { | |
| "epoch": 1.3059033989266546, | |
| "grad_norm": 0.606471839392876, | |
| "learning_rate": 3.910266853206058e-05, | |
| "loss": 0.3227, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3107641935348511, | |
| "step": 730, | |
| "valid_targets_mean": 3896.6, | |
| "valid_targets_min": 545 | |
| }, | |
| { | |
| "epoch": 1.3148479427549196, | |
| "grad_norm": 0.6158261916551597, | |
| "learning_rate": 3.907605239870342e-05, | |
| "loss": 0.3187, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.31847408413887024, | |
| "step": 735, | |
| "valid_targets_mean": 3804.2, | |
| "valid_targets_min": 741 | |
| }, | |
| { | |
| "epoch": 1.3237924865831843, | |
| "grad_norm": 0.6321605305603367, | |
| "learning_rate": 3.904905660478072e-05, | |
| "loss": 0.3249, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3120426833629608, | |
| "step": 740, | |
| "valid_targets_mean": 3962.6, | |
| "valid_targets_min": 1180 | |
| }, | |
| { | |
| "epoch": 1.332737030411449, | |
| "grad_norm": 0.5592948617131595, | |
| "learning_rate": 3.9021681687575465e-05, | |
| "loss": 0.3179, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.323030024766922, | |
| "step": 745, | |
| "valid_targets_mean": 4431.0, | |
| "valid_targets_min": 1535 | |
| }, | |
| { | |
| "epoch": 1.341681574239714, | |
| "grad_norm": 0.5325276240664678, | |
| "learning_rate": 3.8993928191916134e-05, | |
| "loss": 0.3142, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.30110353231430054, | |
| "step": 750, | |
| "valid_targets_mean": 4319.6, | |
| "valid_targets_min": 1804 | |
| }, | |
| { | |
| "epoch": 1.3506261180679786, | |
| "grad_norm": 0.5538507843730288, | |
| "learning_rate": 3.8965796670165856e-05, | |
| "loss": 0.3166, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.31126099824905396, | |
| "step": 755, | |
| "valid_targets_mean": 4311.1, | |
| "valid_targets_min": 1641 | |
| }, | |
| { | |
| "epoch": 1.3595706618962433, | |
| "grad_norm": 0.5678960091362133, | |
| "learning_rate": 3.893728768221139e-05, | |
| "loss": 0.3244, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.305950790643692, | |
| "step": 760, | |
| "valid_targets_mean": 3987.8, | |
| "valid_targets_min": 2194 | |
| }, | |
| { | |
| "epoch": 1.368515205724508, | |
| "grad_norm": 0.5888212617780767, | |
| "learning_rate": 3.8908401795452033e-05, | |
| "loss": 0.3307, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3165014684200287, | |
| "step": 765, | |
| "valid_targets_mean": 4065.1, | |
| "valid_targets_min": 1441 | |
| }, | |
| { | |
| "epoch": 1.3774597495527727, | |
| "grad_norm": 0.6376960259257772, | |
| "learning_rate": 3.8879139584788286e-05, | |
| "loss": 0.3235, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3417049050331116, | |
| "step": 770, | |
| "valid_targets_mean": 3551.9, | |
| "valid_targets_min": 1828 | |
| }, | |
| { | |
| "epoch": 1.3864042933810374, | |
| "grad_norm": 0.5627522323734243, | |
| "learning_rate": 3.884950163261042e-05, | |
| "loss": 0.3218, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.32489728927612305, | |
| "step": 775, | |
| "valid_targets_mean": 3681.4, | |
| "valid_targets_min": 1565 | |
| }, | |
| { | |
| "epoch": 1.3953488372093024, | |
| "grad_norm": 0.643663426567113, | |
| "learning_rate": 3.8819488528786904e-05, | |
| "loss": 0.3162, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.29307177662849426, | |
| "step": 780, | |
| "valid_targets_mean": 3725.8, | |
| "valid_targets_min": 778 | |
| }, | |
| { | |
| "epoch": 1.404293381037567, | |
| "grad_norm": 0.5860397601189946, | |
| "learning_rate": 3.878910087065264e-05, | |
| "loss": 0.3255, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.31179744005203247, | |
| "step": 785, | |
| "valid_targets_mean": 3490.3, | |
| "valid_targets_min": 997 | |
| }, | |
| { | |
| "epoch": 1.4132379248658318, | |
| "grad_norm": 0.6006178361895881, | |
| "learning_rate": 3.8758339262997094e-05, | |
| "loss": 0.3207, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3223787546157837, | |
| "step": 790, | |
| "valid_targets_mean": 3681.4, | |
| "valid_targets_min": 1614 | |
| }, | |
| { | |
| "epoch": 1.4221824686940967, | |
| "grad_norm": 0.5703573435024017, | |
| "learning_rate": 3.872720431805224e-05, | |
| "loss": 0.3108, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.31823939085006714, | |
| "step": 795, | |
| "valid_targets_mean": 3783.3, | |
| "valid_targets_min": 474 | |
| }, | |
| { | |
| "epoch": 1.4311270125223614, | |
| "grad_norm": 0.5583054675958945, | |
| "learning_rate": 3.86956966554804e-05, | |
| "loss": 0.3226, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3209839463233948, | |
| "step": 800, | |
| "valid_targets_mean": 4748.8, | |
| "valid_targets_min": 1829 | |
| }, | |
| { | |
| "epoch": 1.4400715563506261, | |
| "grad_norm": 0.5734451464718057, | |
| "learning_rate": 3.8663816902361896e-05, | |
| "loss": 0.3223, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3395638167858124, | |
| "step": 805, | |
| "valid_targets_mean": 4469.4, | |
| "valid_targets_min": 954 | |
| }, | |
| { | |
| "epoch": 1.4490161001788908, | |
| "grad_norm": 0.6240967521849574, | |
| "learning_rate": 3.863156569318256e-05, | |
| "loss": 0.3185, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.342720627784729, | |
| "step": 810, | |
| "valid_targets_mean": 3274.9, | |
| "valid_targets_min": 873 | |
| }, | |
| { | |
| "epoch": 1.4579606440071555, | |
| "grad_norm": 0.5839916094456372, | |
| "learning_rate": 3.8598943669821124e-05, | |
| "loss": 0.3352, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.34921061992645264, | |
| "step": 815, | |
| "valid_targets_mean": 4149.1, | |
| "valid_targets_min": 1776 | |
| }, | |
| { | |
| "epoch": 1.4669051878354205, | |
| "grad_norm": 0.5327258875676051, | |
| "learning_rate": 3.856595148153643e-05, | |
| "loss": 0.332, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.29739612340927124, | |
| "step": 820, | |
| "valid_targets_mean": 4420.4, | |
| "valid_targets_min": 1722 | |
| }, | |
| { | |
| "epoch": 1.4758497316636852, | |
| "grad_norm": 0.5726474256279158, | |
| "learning_rate": 3.853258978495454e-05, | |
| "loss": 0.3283, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3427349925041199, | |
| "step": 825, | |
| "valid_targets_mean": 4086.2, | |
| "valid_targets_min": 754 | |
| }, | |
| { | |
| "epoch": 1.4847942754919499, | |
| "grad_norm": 0.5586681064868766, | |
| "learning_rate": 3.8498859244055616e-05, | |
| "loss": 0.3201, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.30017322301864624, | |
| "step": 830, | |
| "valid_targets_mean": 4213.1, | |
| "valid_targets_min": 1521 | |
| }, | |
| { | |
| "epoch": 1.4937388193202148, | |
| "grad_norm": 0.5545114860278723, | |
| "learning_rate": 3.8464760530160756e-05, | |
| "loss": 0.3041, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3249306082725525, | |
| "step": 835, | |
| "valid_targets_mean": 4479.8, | |
| "valid_targets_min": 1882 | |
| }, | |
| { | |
| "epoch": 1.5026833631484795, | |
| "grad_norm": 0.5851321941617261, | |
| "learning_rate": 3.843029432191858e-05, | |
| "loss": 0.32, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3026468753814697, | |
| "step": 840, | |
| "valid_targets_mean": 3582.4, | |
| "valid_targets_min": 1779 | |
| }, | |
| { | |
| "epoch": 1.5116279069767442, | |
| "grad_norm": 0.6559733463498192, | |
| "learning_rate": 3.839546130529179e-05, | |
| "loss": 0.3174, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.33914104104042053, | |
| "step": 845, | |
| "valid_targets_mean": 4173.3, | |
| "valid_targets_min": 1823 | |
| }, | |
| { | |
| "epoch": 1.520572450805009, | |
| "grad_norm": 0.6371050604836407, | |
| "learning_rate": 3.8360262173543467e-05, | |
| "loss": 0.3255, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.32565081119537354, | |
| "step": 850, | |
| "valid_targets_mean": 3549.8, | |
| "valid_targets_min": 822 | |
| }, | |
| { | |
| "epoch": 1.5295169946332736, | |
| "grad_norm": 0.5768187274321938, | |
| "learning_rate": 3.8324697627223263e-05, | |
| "loss": 0.317, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3176955282688141, | |
| "step": 855, | |
| "valid_targets_mean": 3880.8, | |
| "valid_targets_min": 671 | |
| }, | |
| { | |
| "epoch": 1.5384615384615383, | |
| "grad_norm": 0.5482702475809518, | |
| "learning_rate": 3.828876837415352e-05, | |
| "loss": 0.3107, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2995319962501526, | |
| "step": 860, | |
| "valid_targets_mean": 4659.9, | |
| "valid_targets_min": 1931 | |
| }, | |
| { | |
| "epoch": 1.5474060822898033, | |
| "grad_norm": 0.6038821934595454, | |
| "learning_rate": 3.8252475129415127e-05, | |
| "loss": 0.3161, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.32784345746040344, | |
| "step": 865, | |
| "valid_targets_mean": 3603.8, | |
| "valid_targets_min": 747 | |
| }, | |
| { | |
| "epoch": 1.556350626118068, | |
| "grad_norm": 0.5283881224885371, | |
| "learning_rate": 3.82158186153333e-05, | |
| "loss": 0.3159, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.30888307094573975, | |
| "step": 870, | |
| "valid_targets_mean": 4141.3, | |
| "valid_targets_min": 1381 | |
| }, | |
| { | |
| "epoch": 1.5652951699463329, | |
| "grad_norm": 0.5983795772826802, | |
| "learning_rate": 3.817879956146323e-05, | |
| "loss": 0.3294, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.34233012795448303, | |
| "step": 875, | |
| "valid_targets_mean": 4115.4, | |
| "valid_targets_min": 1008 | |
| }, | |
| { | |
| "epoch": 1.5742397137745976, | |
| "grad_norm": 0.6611039790326494, | |
| "learning_rate": 3.814141870457553e-05, | |
| "loss": 0.3115, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3333245813846588, | |
| "step": 880, | |
| "valid_targets_mean": 3703.6, | |
| "valid_targets_min": 848 | |
| }, | |
| { | |
| "epoch": 1.5831842576028623, | |
| "grad_norm": 0.5991870251745893, | |
| "learning_rate": 3.810367678864159e-05, | |
| "loss": 0.3237, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.30103540420532227, | |
| "step": 885, | |
| "valid_targets_mean": 3587.0, | |
| "valid_targets_min": 1491 | |
| }, | |
| { | |
| "epoch": 1.592128801431127, | |
| "grad_norm": 0.5618095876534, | |
| "learning_rate": 3.806557456481878e-05, | |
| "loss": 0.3343, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.31948399543762207, | |
| "step": 890, | |
| "valid_targets_mean": 3862.9, | |
| "valid_targets_min": 715 | |
| }, | |
| { | |
| "epoch": 1.6010733452593917, | |
| "grad_norm": 0.6334383283680136, | |
| "learning_rate": 3.8027112791435466e-05, | |
| "loss": 0.3288, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.31764253973960876, | |
| "step": 895, | |
| "valid_targets_mean": 3225.9, | |
| "valid_targets_min": 607 | |
| }, | |
| { | |
| "epoch": 1.6100178890876564, | |
| "grad_norm": 0.582112973475375, | |
| "learning_rate": 3.7988292233975947e-05, | |
| "loss": 0.3329, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.33960235118865967, | |
| "step": 900, | |
| "valid_targets_mean": 4268.9, | |
| "valid_targets_min": 1483 | |
| }, | |
| { | |
| "epoch": 1.6189624329159211, | |
| "grad_norm": 0.5815635189350303, | |
| "learning_rate": 3.7949113665065226e-05, | |
| "loss": 0.3234, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3374456763267517, | |
| "step": 905, | |
| "valid_targets_mean": 4174.2, | |
| "valid_targets_min": 687 | |
| }, | |
| { | |
| "epoch": 1.627906976744186, | |
| "grad_norm": 1.0497579746341565, | |
| "learning_rate": 3.7909577864453593e-05, | |
| "loss": 0.325, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3294409513473511, | |
| "step": 910, | |
| "valid_targets_mean": 4344.9, | |
| "valid_targets_min": 827 | |
| }, | |
| { | |
| "epoch": 1.6368515205724508, | |
| "grad_norm": 0.567491242330487, | |
| "learning_rate": 3.786968561900116e-05, | |
| "loss": 0.3105, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2935563325881958, | |
| "step": 915, | |
| "valid_targets_mean": 3874.4, | |
| "valid_targets_min": 1672 | |
| }, | |
| { | |
| "epoch": 1.6457960644007157, | |
| "grad_norm": 0.6318743528248707, | |
| "learning_rate": 3.782943772266213e-05, | |
| "loss": 0.3191, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3036611080169678, | |
| "step": 920, | |
| "valid_targets_mean": 3250.3, | |
| "valid_targets_min": 911 | |
| }, | |
| { | |
| "epoch": 1.6547406082289804, | |
| "grad_norm": 0.6392903008216162, | |
| "learning_rate": 3.7788834976469095e-05, | |
| "loss": 0.3083, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3205622434616089, | |
| "step": 925, | |
| "valid_targets_mean": 3400.8, | |
| "valid_targets_min": 978 | |
| }, | |
| { | |
| "epoch": 1.663685152057245, | |
| "grad_norm": 0.5491612081779448, | |
| "learning_rate": 3.7747878188516965e-05, | |
| "loss": 0.3137, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3106827437877655, | |
| "step": 930, | |
| "valid_targets_mean": 4198.3, | |
| "valid_targets_min": 1757 | |
| }, | |
| { | |
| "epoch": 1.6726296958855098, | |
| "grad_norm": 0.5396436465197288, | |
| "learning_rate": 3.770656817394703e-05, | |
| "loss": 0.3168, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.30610522627830505, | |
| "step": 935, | |
| "valid_targets_mean": 4018.6, | |
| "valid_targets_min": 1540 | |
| }, | |
| { | |
| "epoch": 1.6815742397137745, | |
| "grad_norm": 0.4890047892792955, | |
| "learning_rate": 3.7664905754930616e-05, | |
| "loss": 0.3155, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.31638064980506897, | |
| "step": 940, | |
| "valid_targets_mean": 4828.7, | |
| "valid_targets_min": 1454 | |
| }, | |
| { | |
| "epoch": 1.6905187835420392, | |
| "grad_norm": 0.5488512250424282, | |
| "learning_rate": 3.762289176065276e-05, | |
| "loss": 0.3167, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3114585280418396, | |
| "step": 945, | |
| "valid_targets_mean": 4037.7, | |
| "valid_targets_min": 600 | |
| }, | |
| { | |
| "epoch": 1.6994633273703041, | |
| "grad_norm": 0.643517620390725, | |
| "learning_rate": 3.758052702729576e-05, | |
| "loss": 0.3127, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.305495023727417, | |
| "step": 950, | |
| "valid_targets_mean": 4262.5, | |
| "valid_targets_min": 1184 | |
| }, | |
| { | |
| "epoch": 1.7084078711985689, | |
| "grad_norm": 0.5457086631910586, | |
| "learning_rate": 3.753781239802245e-05, | |
| "loss": 0.3132, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2907237410545349, | |
| "step": 955, | |
| "valid_targets_mean": 3675.6, | |
| "valid_targets_min": 1495 | |
| }, | |
| { | |
| "epoch": 1.7173524150268338, | |
| "grad_norm": 0.5241171467893824, | |
| "learning_rate": 3.749474872295946e-05, | |
| "loss": 0.3125, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2793015241622925, | |
| "step": 960, | |
| "valid_targets_mean": 3477.6, | |
| "valid_targets_min": 1785 | |
| }, | |
| { | |
| "epoch": 1.7262969588550985, | |
| "grad_norm": 0.5513650345883626, | |
| "learning_rate": 3.745133685918032e-05, | |
| "loss": 0.321, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3557593822479248, | |
| "step": 965, | |
| "valid_targets_mean": 4640.1, | |
| "valid_targets_min": 2109 | |
| }, | |
| { | |
| "epoch": 1.7352415026833632, | |
| "grad_norm": 0.526841550249395, | |
| "learning_rate": 3.740757767068834e-05, | |
| "loss": 0.3022, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2938116788864136, | |
| "step": 970, | |
| "valid_targets_mean": 4656.8, | |
| "valid_targets_min": 1424 | |
| }, | |
| { | |
| "epoch": 1.744186046511628, | |
| "grad_norm": 0.6055246993537275, | |
| "learning_rate": 3.7363472028399476e-05, | |
| "loss": 0.3248, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3154802918434143, | |
| "step": 975, | |
| "valid_targets_mean": 3483.9, | |
| "valid_targets_min": 1692 | |
| }, | |
| { | |
| "epoch": 1.7531305903398926, | |
| "grad_norm": 0.603322342815491, | |
| "learning_rate": 3.7319020810124965e-05, | |
| "loss": 0.3021, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3003346920013428, | |
| "step": 980, | |
| "valid_targets_mean": 3828.6, | |
| "valid_targets_min": 621 | |
| }, | |
| { | |
| "epoch": 1.7620751341681573, | |
| "grad_norm": 0.5626166447337553, | |
| "learning_rate": 3.727422490055386e-05, | |
| "loss": 0.3148, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3034425973892212, | |
| "step": 985, | |
| "valid_targets_mean": 4099.6, | |
| "valid_targets_min": 836 | |
| }, | |
| { | |
| "epoch": 1.7710196779964222, | |
| "grad_norm": 0.4590848854306322, | |
| "learning_rate": 3.72290851912354e-05, | |
| "loss": 0.314, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3024485111236572, | |
| "step": 990, | |
| "valid_targets_mean": 5403.2, | |
| "valid_targets_min": 2338 | |
| }, | |
| { | |
| "epoch": 1.779964221824687, | |
| "grad_norm": 0.5634666645265822, | |
| "learning_rate": 3.718360258056133e-05, | |
| "loss": 0.3173, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.30547595024108887, | |
| "step": 995, | |
| "valid_targets_mean": 3841.9, | |
| "valid_targets_min": 1685 | |
| }, | |
| { | |
| "epoch": 1.7889087656529516, | |
| "grad_norm": 0.5618746894535553, | |
| "learning_rate": 3.713777797374794e-05, | |
| "loss": 0.322, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.33232662081718445, | |
| "step": 1000, | |
| "valid_targets_mean": 3674.4, | |
| "valid_targets_min": 663 | |
| }, | |
| { | |
| "epoch": 1.7978533094812166, | |
| "grad_norm": 0.6550093924031264, | |
| "learning_rate": 3.709161228281811e-05, | |
| "loss": 0.3142, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.30034154653549194, | |
| "step": 1005, | |
| "valid_targets_mean": 3396.8, | |
| "valid_targets_min": 1109 | |
| }, | |
| { | |
| "epoch": 1.8067978533094813, | |
| "grad_norm": 0.6073692572388313, | |
| "learning_rate": 3.704510642658314e-05, | |
| "loss": 0.3203, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3542916178703308, | |
| "step": 1010, | |
| "valid_targets_mean": 3448.8, | |
| "valid_targets_min": 795 | |
| }, | |
| { | |
| "epoch": 1.815742397137746, | |
| "grad_norm": 0.6171396796866804, | |
| "learning_rate": 3.699826133062443e-05, | |
| "loss": 0.3118, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3225131630897522, | |
| "step": 1015, | |
| "valid_targets_mean": 3759.8, | |
| "valid_targets_min": 1114 | |
| }, | |
| { | |
| "epoch": 1.8246869409660107, | |
| "grad_norm": 0.6099863228762299, | |
| "learning_rate": 3.6951077927275126e-05, | |
| "loss": 0.332, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3298446834087372, | |
| "step": 1020, | |
| "valid_targets_mean": 3864.5, | |
| "valid_targets_min": 1052 | |
| }, | |
| { | |
| "epoch": 1.8336314847942754, | |
| "grad_norm": 0.5524141402638945, | |
| "learning_rate": 3.6903557155601503e-05, | |
| "loss": 0.3117, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.291124165058136, | |
| "step": 1025, | |
| "valid_targets_mean": 3606.2, | |
| "valid_targets_min": 1973 | |
| }, | |
| { | |
| "epoch": 1.84257602862254, | |
| "grad_norm": 0.5169690200399784, | |
| "learning_rate": 3.685569996138431e-05, | |
| "loss": 0.3065, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3137836456298828, | |
| "step": 1030, | |
| "valid_targets_mean": 4619.6, | |
| "valid_targets_min": 1219 | |
| }, | |
| { | |
| "epoch": 1.851520572450805, | |
| "grad_norm": 0.5316610746415315, | |
| "learning_rate": 3.680750729709993e-05, | |
| "loss": 0.3173, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.33356547355651855, | |
| "step": 1035, | |
| "valid_targets_mean": 4652.9, | |
| "valid_targets_min": 981 | |
| }, | |
| { | |
| "epoch": 1.8604651162790697, | |
| "grad_norm": 0.5559235856424838, | |
| "learning_rate": 3.675898012190143e-05, | |
| "loss": 0.3105, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.31008175015449524, | |
| "step": 1040, | |
| "valid_targets_mean": 3755.5, | |
| "valid_targets_min": 1368 | |
| }, | |
| { | |
| "epoch": 1.8694096601073347, | |
| "grad_norm": 0.6298206469639374, | |
| "learning_rate": 3.6710119401599474e-05, | |
| "loss": 0.3252, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.33441591262817383, | |
| "step": 1045, | |
| "valid_targets_mean": 3854.9, | |
| "valid_targets_min": 1316 | |
| }, | |
| { | |
| "epoch": 1.8783542039355994, | |
| "grad_norm": 0.5389022768825737, | |
| "learning_rate": 3.6660926108643086e-05, | |
| "loss": 0.3179, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3052956461906433, | |
| "step": 1050, | |
| "valid_targets_mean": 3938.4, | |
| "valid_targets_min": 680 | |
| }, | |
| { | |
| "epoch": 1.887298747763864, | |
| "grad_norm": 0.5801147385776886, | |
| "learning_rate": 3.661140122210032e-05, | |
| "loss": 0.309, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.32204878330230713, | |
| "step": 1055, | |
| "valid_targets_mean": 3663.2, | |
| "valid_targets_min": 839 | |
| }, | |
| { | |
| "epoch": 1.8962432915921288, | |
| "grad_norm": 0.5606779533763502, | |
| "learning_rate": 3.656154572763877e-05, | |
| "loss": 0.3211, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3348532021045685, | |
| "step": 1060, | |
| "valid_targets_mean": 3548.2, | |
| "valid_targets_min": 622 | |
| }, | |
| { | |
| "epoch": 1.9051878354203935, | |
| "grad_norm": 0.5573332010454619, | |
| "learning_rate": 3.651136061750592e-05, | |
| "loss": 0.3252, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3448393940925598, | |
| "step": 1065, | |
| "valid_targets_mean": 3946.6, | |
| "valid_targets_min": 748 | |
| }, | |
| { | |
| "epoch": 1.9141323792486582, | |
| "grad_norm": 0.5490778045530725, | |
| "learning_rate": 3.646084689050942e-05, | |
| "loss": 0.3189, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3161720633506775, | |
| "step": 1070, | |
| "valid_targets_mean": 3859.4, | |
| "valid_targets_min": 751 | |
| }, | |
| { | |
| "epoch": 1.9230769230769231, | |
| "grad_norm": 0.5894163589977383, | |
| "learning_rate": 3.641000555199725e-05, | |
| "loss": 0.3105, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3114989697933197, | |
| "step": 1075, | |
| "valid_targets_mean": 3375.2, | |
| "valid_targets_min": 710 | |
| }, | |
| { | |
| "epoch": 1.9320214669051878, | |
| "grad_norm": 0.5349889235476035, | |
| "learning_rate": 3.6358837613837604e-05, | |
| "loss": 0.315, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.30235400795936584, | |
| "step": 1080, | |
| "valid_targets_mean": 3989.0, | |
| "valid_targets_min": 1691 | |
| }, | |
| { | |
| "epoch": 1.9409660107334525, | |
| "grad_norm": 0.6736523490222415, | |
| "learning_rate": 3.630734409439887e-05, | |
| "loss": 0.3227, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3229427635669708, | |
| "step": 1085, | |
| "valid_targets_mean": 3905.6, | |
| "valid_targets_min": 950 | |
| }, | |
| { | |
| "epoch": 1.9499105545617175, | |
| "grad_norm": 0.5727534078817972, | |
| "learning_rate": 3.625552601852928e-05, | |
| "loss": 0.3119, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.31192246079444885, | |
| "step": 1090, | |
| "valid_targets_mean": 3771.9, | |
| "valid_targets_min": 658 | |
| }, | |
| { | |
| "epoch": 1.9588550983899822, | |
| "grad_norm": 0.6120985202643506, | |
| "learning_rate": 3.6203384417536566e-05, | |
| "loss": 0.31, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3371928036212921, | |
| "step": 1095, | |
| "valid_targets_mean": 3940.8, | |
| "valid_targets_min": 1372 | |
| }, | |
| { | |
| "epoch": 1.9677996422182469, | |
| "grad_norm": 0.5549601857245118, | |
| "learning_rate": 3.615092032916736e-05, | |
| "loss": 0.3106, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3066266179084778, | |
| "step": 1100, | |
| "valid_targets_mean": 3876.9, | |
| "valid_targets_min": 1122 | |
| }, | |
| { | |
| "epoch": 1.9767441860465116, | |
| "grad_norm": 0.5131380359563835, | |
| "learning_rate": 3.6098134797586646e-05, | |
| "loss": 0.3019, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2861686646938324, | |
| "step": 1105, | |
| "valid_targets_mean": 4285.8, | |
| "valid_targets_min": 1706 | |
| }, | |
| { | |
| "epoch": 1.9856887298747763, | |
| "grad_norm": 0.527229318850269, | |
| "learning_rate": 3.604502887335688e-05, | |
| "loss": 0.3178, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.317037433385849, | |
| "step": 1110, | |
| "valid_targets_mean": 4532.9, | |
| "valid_targets_min": 1358 | |
| }, | |
| { | |
| "epoch": 1.994633273703041, | |
| "grad_norm": 0.6132587433714716, | |
| "learning_rate": 3.599160361341715e-05, | |
| "loss": 0.3246, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.35272109508514404, | |
| "step": 1115, | |
| "valid_targets_mean": 4244.4, | |
| "valid_targets_min": 1862 | |
| }, | |
| { | |
| "epoch": 2.0035778175313057, | |
| "grad_norm": 0.5593406465085026, | |
| "learning_rate": 3.59378600810621e-05, | |
| "loss": 0.3091, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.30550044775009155, | |
| "step": 1120, | |
| "valid_targets_mean": 3629.8, | |
| "valid_targets_min": 1287 | |
| }, | |
| { | |
| "epoch": 2.012522361359571, | |
| "grad_norm": 0.5559035299742234, | |
| "learning_rate": 3.588379934592078e-05, | |
| "loss": 0.3007, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2961064577102661, | |
| "step": 1125, | |
| "valid_targets_mean": 4219.7, | |
| "valid_targets_min": 2082 | |
| }, | |
| { | |
| "epoch": 2.0214669051878356, | |
| "grad_norm": 0.5222588666049305, | |
| "learning_rate": 3.5829422483935374e-05, | |
| "loss": 0.2991, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2910686135292053, | |
| "step": 1130, | |
| "valid_targets_mean": 4635.5, | |
| "valid_targets_min": 720 | |
| }, | |
| { | |
| "epoch": 2.0304114490161003, | |
| "grad_norm": 0.5492856936529431, | |
| "learning_rate": 3.577473057733975e-05, | |
| "loss": 0.2912, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2696704864501953, | |
| "step": 1135, | |
| "valid_targets_mean": 3424.2, | |
| "valid_targets_min": 1441 | |
| }, | |
| { | |
| "epoch": 2.039355992844365, | |
| "grad_norm": 0.5805919255184491, | |
| "learning_rate": 3.571972471463795e-05, | |
| "loss": 0.2914, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2642717957496643, | |
| "step": 1140, | |
| "valid_targets_mean": 3726.3, | |
| "valid_targets_min": 1936 | |
| }, | |
| { | |
| "epoch": 2.0483005366726297, | |
| "grad_norm": 0.5654677847464393, | |
| "learning_rate": 3.566440599058253e-05, | |
| "loss": 0.2984, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.31323423981666565, | |
| "step": 1145, | |
| "valid_targets_mean": 4155.1, | |
| "valid_targets_min": 1574 | |
| }, | |
| { | |
| "epoch": 2.0572450805008944, | |
| "grad_norm": 0.6258943561807471, | |
| "learning_rate": 3.560877550615275e-05, | |
| "loss": 0.2852, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2951895594596863, | |
| "step": 1150, | |
| "valid_targets_mean": 3085.2, | |
| "valid_targets_min": 732 | |
| }, | |
| { | |
| "epoch": 2.066189624329159, | |
| "grad_norm": 0.48185508119836556, | |
| "learning_rate": 3.555283436853267e-05, | |
| "loss": 0.292, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.277047336101532, | |
| "step": 1155, | |
| "valid_targets_mean": 4333.4, | |
| "valid_targets_min": 1674 | |
| }, | |
| { | |
| "epoch": 2.075134168157424, | |
| "grad_norm": 0.5337477447807386, | |
| "learning_rate": 3.549658369108911e-05, | |
| "loss": 0.2844, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.290058970451355, | |
| "step": 1160, | |
| "valid_targets_mean": 4003.7, | |
| "valid_targets_min": 667 | |
| }, | |
| { | |
| "epoch": 2.084078711985689, | |
| "grad_norm": 0.573076273715761, | |
| "learning_rate": 3.544002459334952e-05, | |
| "loss": 0.2941, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2986866235733032, | |
| "step": 1165, | |
| "valid_targets_mean": 3509.6, | |
| "valid_targets_min": 827 | |
| }, | |
| { | |
| "epoch": 2.0930232558139537, | |
| "grad_norm": 0.6434121744773865, | |
| "learning_rate": 3.5383158200979636e-05, | |
| "loss": 0.3013, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.29687678813934326, | |
| "step": 1170, | |
| "valid_targets_mean": 2894.1, | |
| "valid_targets_min": 1532 | |
| }, | |
| { | |
| "epoch": 2.1019677996422184, | |
| "grad_norm": 1.0345078244238066, | |
| "learning_rate": 3.532598564576117e-05, | |
| "loss": 0.2923, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27670419216156006, | |
| "step": 1175, | |
| "valid_targets_mean": 3932.2, | |
| "valid_targets_min": 798 | |
| }, | |
| { | |
| "epoch": 2.110912343470483, | |
| "grad_norm": 0.5405380881490202, | |
| "learning_rate": 3.526850806556919e-05, | |
| "loss": 0.302, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2855556011199951, | |
| "step": 1180, | |
| "valid_targets_mean": 3778.3, | |
| "valid_targets_min": 696 | |
| }, | |
| { | |
| "epoch": 2.1198568872987478, | |
| "grad_norm": 0.5976648828621719, | |
| "learning_rate": 3.521072660434952e-05, | |
| "loss": 0.2899, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28265494108200073, | |
| "step": 1185, | |
| "valid_targets_mean": 3455.8, | |
| "valid_targets_min": 866 | |
| }, | |
| { | |
| "epoch": 2.1288014311270125, | |
| "grad_norm": 0.518386120621355, | |
| "learning_rate": 3.5152642412095984e-05, | |
| "loss": 0.2869, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2857966721057892, | |
| "step": 1190, | |
| "valid_targets_mean": 4677.8, | |
| "valid_targets_min": 2305 | |
| }, | |
| { | |
| "epoch": 2.137745974955277, | |
| "grad_norm": 0.5447094720161839, | |
| "learning_rate": 3.5094256644827474e-05, | |
| "loss": 0.3016, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3032097816467285, | |
| "step": 1195, | |
| "valid_targets_mean": 4234.6, | |
| "valid_targets_min": 1055 | |
| }, | |
| { | |
| "epoch": 2.146690518783542, | |
| "grad_norm": 0.5406564490251188, | |
| "learning_rate": 3.503557046456501e-05, | |
| "loss": 0.2806, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2764890193939209, | |
| "step": 1200, | |
| "valid_targets_mean": 4025.9, | |
| "valid_targets_min": 1610 | |
| }, | |
| { | |
| "epoch": 2.1556350626118066, | |
| "grad_norm": 0.5167292603339387, | |
| "learning_rate": 3.4976585039308535e-05, | |
| "loss": 0.3053, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2947821617126465, | |
| "step": 1205, | |
| "valid_targets_mean": 4129.1, | |
| "valid_targets_min": 1605 | |
| }, | |
| { | |
| "epoch": 2.1645796064400717, | |
| "grad_norm": 0.7469544699185315, | |
| "learning_rate": 3.491730154301372e-05, | |
| "loss": 0.3068, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3116210401058197, | |
| "step": 1210, | |
| "valid_targets_mean": 3251.5, | |
| "valid_targets_min": 1606 | |
| }, | |
| { | |
| "epoch": 2.1735241502683365, | |
| "grad_norm": 0.5769662832718492, | |
| "learning_rate": 3.485772115556859e-05, | |
| "loss": 0.2967, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2989100217819214, | |
| "step": 1215, | |
| "valid_targets_mean": 3717.8, | |
| "valid_targets_min": 1327 | |
| }, | |
| { | |
| "epoch": 2.182468694096601, | |
| "grad_norm": 0.5888498917346773, | |
| "learning_rate": 3.4797845062770045e-05, | |
| "loss": 0.2985, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2971697747707367, | |
| "step": 1220, | |
| "valid_targets_mean": 3346.9, | |
| "valid_targets_min": 1827 | |
| }, | |
| { | |
| "epoch": 2.191413237924866, | |
| "grad_norm": 0.5385601523875473, | |
| "learning_rate": 3.473767445630022e-05, | |
| "loss": 0.2938, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2714182436466217, | |
| "step": 1225, | |
| "valid_targets_mean": 3834.8, | |
| "valid_targets_min": 1393 | |
| }, | |
| { | |
| "epoch": 2.2003577817531306, | |
| "grad_norm": 0.5095146941186255, | |
| "learning_rate": 3.467721053370284e-05, | |
| "loss": 0.2914, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3141205608844757, | |
| "step": 1230, | |
| "valid_targets_mean": 4676.5, | |
| "valid_targets_min": 1372 | |
| }, | |
| { | |
| "epoch": 2.2093023255813953, | |
| "grad_norm": 0.6020826429693371, | |
| "learning_rate": 3.4616454498359306e-05, | |
| "loss": 0.3047, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.29343247413635254, | |
| "step": 1235, | |
| "valid_targets_mean": 3196.5, | |
| "valid_targets_min": 1524 | |
| }, | |
| { | |
| "epoch": 2.21824686940966, | |
| "grad_norm": 0.5403496852716332, | |
| "learning_rate": 3.4555407559464825e-05, | |
| "loss": 0.3081, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3097769618034363, | |
| "step": 1240, | |
| "valid_targets_mean": 4774.0, | |
| "valid_targets_min": 1541 | |
| }, | |
| { | |
| "epoch": 2.2271914132379247, | |
| "grad_norm": 0.5120840671547888, | |
| "learning_rate": 3.4494070932004274e-05, | |
| "loss": 0.3015, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.284509539604187, | |
| "step": 1245, | |
| "valid_targets_mean": 4208.5, | |
| "valid_targets_min": 1295 | |
| }, | |
| { | |
| "epoch": 2.23613595706619, | |
| "grad_norm": 0.5914343254506964, | |
| "learning_rate": 3.4432445836728055e-05, | |
| "loss": 0.2834, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2789539694786072, | |
| "step": 1250, | |
| "valid_targets_mean": 4127.3, | |
| "valid_targets_min": 872 | |
| }, | |
| { | |
| "epoch": 2.2450805008944545, | |
| "grad_norm": 0.5645415624490583, | |
| "learning_rate": 3.4370533500127794e-05, | |
| "loss": 0.3038, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.30053240060806274, | |
| "step": 1255, | |
| "valid_targets_mean": 3852.9, | |
| "valid_targets_min": 1580 | |
| }, | |
| { | |
| "epoch": 2.2540250447227193, | |
| "grad_norm": 0.5806828746585898, | |
| "learning_rate": 3.430833515441193e-05, | |
| "loss": 0.3006, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.31069332361221313, | |
| "step": 1260, | |
| "valid_targets_mean": 3907.2, | |
| "valid_targets_min": 1436 | |
| }, | |
| { | |
| "epoch": 2.262969588550984, | |
| "grad_norm": 0.5626831319283796, | |
| "learning_rate": 3.424585203748119e-05, | |
| "loss": 0.3021, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3043138086795807, | |
| "step": 1265, | |
| "valid_targets_mean": 4106.3, | |
| "valid_targets_min": 1718 | |
| }, | |
| { | |
| "epoch": 2.2719141323792487, | |
| "grad_norm": 0.5419828214534085, | |
| "learning_rate": 3.4183085392903965e-05, | |
| "loss": 0.285, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28403908014297485, | |
| "step": 1270, | |
| "valid_targets_mean": 3841.6, | |
| "valid_targets_min": 1794 | |
| }, | |
| { | |
| "epoch": 2.2808586762075134, | |
| "grad_norm": 0.5881372643406569, | |
| "learning_rate": 3.41200364698915e-05, | |
| "loss": 0.2899, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.30241650342941284, | |
| "step": 1275, | |
| "valid_targets_mean": 3480.4, | |
| "valid_targets_min": 1784 | |
| }, | |
| { | |
| "epoch": 2.289803220035778, | |
| "grad_norm": 0.5027336153791081, | |
| "learning_rate": 3.405670652327313e-05, | |
| "loss": 0.3028, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28809165954589844, | |
| "step": 1280, | |
| "valid_targets_mean": 4556.8, | |
| "valid_targets_min": 2196 | |
| }, | |
| { | |
| "epoch": 2.298747763864043, | |
| "grad_norm": 0.5959906299682153, | |
| "learning_rate": 3.399309681347123e-05, | |
| "loss": 0.2956, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.305059552192688, | |
| "step": 1285, | |
| "valid_targets_mean": 4217.6, | |
| "valid_targets_min": 1517 | |
| }, | |
| { | |
| "epoch": 2.3076923076923075, | |
| "grad_norm": 0.47359707150452485, | |
| "learning_rate": 3.392920860647617e-05, | |
| "loss": 0.2975, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3012004792690277, | |
| "step": 1290, | |
| "valid_targets_mean": 4578.0, | |
| "valid_targets_min": 1297 | |
| }, | |
| { | |
| "epoch": 2.3166368515205726, | |
| "grad_norm": 0.5369785466687547, | |
| "learning_rate": 3.3865043173821074e-05, | |
| "loss": 0.2854, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2727619409561157, | |
| "step": 1295, | |
| "valid_targets_mean": 3760.1, | |
| "valid_targets_min": 1539 | |
| }, | |
| { | |
| "epoch": 2.3255813953488373, | |
| "grad_norm": 0.57676869627547, | |
| "learning_rate": 3.380060179255656e-05, | |
| "loss": 0.3036, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.31669068336486816, | |
| "step": 1300, | |
| "valid_targets_mean": 3117.1, | |
| "valid_targets_min": 1662 | |
| }, | |
| { | |
| "epoch": 2.334525939177102, | |
| "grad_norm": 0.5283474540373422, | |
| "learning_rate": 3.37358857452253e-05, | |
| "loss": 0.3005, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.303724467754364, | |
| "step": 1305, | |
| "valid_targets_mean": 4069.7, | |
| "valid_targets_min": 1847 | |
| }, | |
| { | |
| "epoch": 2.3434704830053668, | |
| "grad_norm": 0.48205035846371097, | |
| "learning_rate": 3.367089631983651e-05, | |
| "loss": 0.3018, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27161988615989685, | |
| "step": 1310, | |
| "valid_targets_mean": 4610.1, | |
| "valid_targets_min": 625 | |
| }, | |
| { | |
| "epoch": 2.3524150268336315, | |
| "grad_norm": 0.5018852845147292, | |
| "learning_rate": 3.360563480984029e-05, | |
| "loss": 0.2908, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2748308479785919, | |
| "step": 1315, | |
| "valid_targets_mean": 4035.2, | |
| "valid_targets_min": 1483 | |
| }, | |
| { | |
| "epoch": 2.361359570661896, | |
| "grad_norm": 0.6082741514089518, | |
| "learning_rate": 3.3540102514101904e-05, | |
| "loss": 0.3009, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2947337031364441, | |
| "step": 1320, | |
| "valid_targets_mean": 3258.3, | |
| "valid_targets_min": 863 | |
| }, | |
| { | |
| "epoch": 2.370304114490161, | |
| "grad_norm": 0.5119247145544926, | |
| "learning_rate": 3.347430073687592e-05, | |
| "loss": 0.3057, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3057720363140106, | |
| "step": 1325, | |
| "valid_targets_mean": 5325.2, | |
| "valid_targets_min": 2447 | |
| }, | |
| { | |
| "epoch": 2.3792486583184256, | |
| "grad_norm": 0.5545132272446245, | |
| "learning_rate": 3.340823078778024e-05, | |
| "loss": 0.2904, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.329643577337265, | |
| "step": 1330, | |
| "valid_targets_mean": 4703.9, | |
| "valid_targets_min": 822 | |
| }, | |
| { | |
| "epoch": 2.3881932021466907, | |
| "grad_norm": 0.5611524701506613, | |
| "learning_rate": 3.3341893981770086e-05, | |
| "loss": 0.2999, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28520113229751587, | |
| "step": 1335, | |
| "valid_targets_mean": 3347.0, | |
| "valid_targets_min": 1432 | |
| }, | |
| { | |
| "epoch": 2.3971377459749554, | |
| "grad_norm": 0.5360100187741607, | |
| "learning_rate": 3.327529163911174e-05, | |
| "loss": 0.29, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27396950125694275, | |
| "step": 1340, | |
| "valid_targets_mean": 4146.1, | |
| "valid_targets_min": 1777 | |
| }, | |
| { | |
| "epoch": 2.40608228980322, | |
| "grad_norm": 0.5232182650009217, | |
| "learning_rate": 3.320842508535636e-05, | |
| "loss": 0.3023, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.294733464717865, | |
| "step": 1345, | |
| "valid_targets_mean": 3978.9, | |
| "valid_targets_min": 1969 | |
| }, | |
| { | |
| "epoch": 2.415026833631485, | |
| "grad_norm": 0.6341635562060696, | |
| "learning_rate": 3.314129565131355e-05, | |
| "loss": 0.3001, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2832929491996765, | |
| "step": 1350, | |
| "valid_targets_mean": 3560.1, | |
| "valid_targets_min": 1561 | |
| }, | |
| { | |
| "epoch": 2.4239713774597496, | |
| "grad_norm": 0.5512350224651154, | |
| "learning_rate": 3.3073904673024854e-05, | |
| "loss": 0.3018, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2930479943752289, | |
| "step": 1355, | |
| "valid_targets_mean": 4085.1, | |
| "valid_targets_min": 1762 | |
| }, | |
| { | |
| "epoch": 2.4329159212880143, | |
| "grad_norm": 0.5089586101245678, | |
| "learning_rate": 3.300625349173723e-05, | |
| "loss": 0.2983, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.304643452167511, | |
| "step": 1360, | |
| "valid_targets_mean": 4801.9, | |
| "valid_targets_min": 1725 | |
| }, | |
| { | |
| "epoch": 2.441860465116279, | |
| "grad_norm": 0.5712139371917907, | |
| "learning_rate": 3.29383434538763e-05, | |
| "loss": 0.2893, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.290343314409256, | |
| "step": 1365, | |
| "valid_targets_mean": 4540.5, | |
| "valid_targets_min": 1504 | |
| }, | |
| { | |
| "epoch": 2.4508050089445437, | |
| "grad_norm": 0.6223435319453212, | |
| "learning_rate": 3.287017591101957e-05, | |
| "loss": 0.3061, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3304898738861084, | |
| "step": 1370, | |
| "valid_targets_mean": 3340.2, | |
| "valid_targets_min": 1231 | |
| }, | |
| { | |
| "epoch": 2.4597495527728084, | |
| "grad_norm": 0.548835445285655, | |
| "learning_rate": 3.2801752219869536e-05, | |
| "loss": 0.2962, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.29954493045806885, | |
| "step": 1375, | |
| "valid_targets_mean": 3768.6, | |
| "valid_targets_min": 1507 | |
| }, | |
| { | |
| "epoch": 2.4686940966010735, | |
| "grad_norm": 0.5273857106581554, | |
| "learning_rate": 3.273307374222667e-05, | |
| "loss": 0.2956, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.29336017370224, | |
| "step": 1380, | |
| "valid_targets_mean": 3922.8, | |
| "valid_targets_min": 527 | |
| }, | |
| { | |
| "epoch": 2.4776386404293382, | |
| "grad_norm": 0.461716385493086, | |
| "learning_rate": 3.266414184496233e-05, | |
| "loss": 0.2867, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.29551392793655396, | |
| "step": 1385, | |
| "valid_targets_mean": 4960.4, | |
| "valid_targets_min": 1853 | |
| }, | |
| { | |
| "epoch": 2.486583184257603, | |
| "grad_norm": 0.5233231328137401, | |
| "learning_rate": 3.2594957899991566e-05, | |
| "loss": 0.3022, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.30951058864593506, | |
| "step": 1390, | |
| "valid_targets_mean": 4894.6, | |
| "valid_targets_min": 1705 | |
| }, | |
| { | |
| "epoch": 2.4955277280858676, | |
| "grad_norm": 0.5483474982667876, | |
| "learning_rate": 3.2525523284245766e-05, | |
| "loss": 0.2837, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2956357002258301, | |
| "step": 1395, | |
| "valid_targets_mean": 4093.8, | |
| "valid_targets_min": 1158 | |
| }, | |
| { | |
| "epoch": 2.5044722719141324, | |
| "grad_norm": 0.5470117760522922, | |
| "learning_rate": 3.245583937964532e-05, | |
| "loss": 0.2757, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.29229462146759033, | |
| "step": 1400, | |
| "valid_targets_mean": 4277.4, | |
| "valid_targets_min": 1526 | |
| }, | |
| { | |
| "epoch": 2.513416815742397, | |
| "grad_norm": 0.5692210997559785, | |
| "learning_rate": 3.238590757307206e-05, | |
| "loss": 0.2907, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28669655323028564, | |
| "step": 1405, | |
| "valid_targets_mean": 4107.9, | |
| "valid_targets_min": 2476 | |
| }, | |
| { | |
| "epoch": 2.5223613595706618, | |
| "grad_norm": 0.5999394928447368, | |
| "learning_rate": 3.2315729256341686e-05, | |
| "loss": 0.3037, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3192169666290283, | |
| "step": 1410, | |
| "valid_targets_mean": 3115.0, | |
| "valid_targets_min": 482 | |
| }, | |
| { | |
| "epoch": 2.531305903398927, | |
| "grad_norm": 0.5067438226356301, | |
| "learning_rate": 3.2245305826176063e-05, | |
| "loss": 0.2978, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.30782514810562134, | |
| "step": 1415, | |
| "valid_targets_mean": 4176.4, | |
| "valid_targets_min": 1489 | |
| }, | |
| { | |
| "epoch": 2.5402504472271916, | |
| "grad_norm": 0.5670153939929208, | |
| "learning_rate": 3.217463868417541e-05, | |
| "loss": 0.2904, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2898740768432617, | |
| "step": 1420, | |
| "valid_targets_mean": 4424.1, | |
| "valid_targets_min": 877 | |
| }, | |
| { | |
| "epoch": 2.5491949910554563, | |
| "grad_norm": 0.5320142185017729, | |
| "learning_rate": 3.2103729236790434e-05, | |
| "loss": 0.292, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3135606646537781, | |
| "step": 1425, | |
| "valid_targets_mean": 4807.0, | |
| "valid_targets_min": 1759 | |
| }, | |
| { | |
| "epoch": 2.558139534883721, | |
| "grad_norm": 0.5377685633015429, | |
| "learning_rate": 3.203257889529428e-05, | |
| "loss": 0.3005, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3080807626247406, | |
| "step": 1430, | |
| "valid_targets_mean": 4145.8, | |
| "valid_targets_min": 1470 | |
| }, | |
| { | |
| "epoch": 2.5670840787119857, | |
| "grad_norm": 0.9791809632288329, | |
| "learning_rate": 3.196118907575452e-05, | |
| "loss": 0.3046, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.32376378774642944, | |
| "step": 1435, | |
| "valid_targets_mean": 2997.5, | |
| "valid_targets_min": 1849 | |
| }, | |
| { | |
| "epoch": 2.5760286225402504, | |
| "grad_norm": 0.5199580890989173, | |
| "learning_rate": 3.188956119900491e-05, | |
| "loss": 0.2942, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27378737926483154, | |
| "step": 1440, | |
| "valid_targets_mean": 3816.5, | |
| "valid_targets_min": 835 | |
| }, | |
| { | |
| "epoch": 2.584973166368515, | |
| "grad_norm": 0.5360826743649624, | |
| "learning_rate": 3.181769669061713e-05, | |
| "loss": 0.2917, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3072686791419983, | |
| "step": 1445, | |
| "valid_targets_mean": 3774.3, | |
| "valid_targets_min": 529 | |
| }, | |
| { | |
| "epoch": 2.59391771019678, | |
| "grad_norm": 0.5009124650640604, | |
| "learning_rate": 3.174559698087244e-05, | |
| "loss": 0.3046, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2810012698173523, | |
| "step": 1450, | |
| "valid_targets_mean": 3802.2, | |
| "valid_targets_min": 1494 | |
| }, | |
| { | |
| "epoch": 2.6028622540250446, | |
| "grad_norm": 0.5463511449260107, | |
| "learning_rate": 3.1673263504733136e-05, | |
| "loss": 0.2966, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28489333391189575, | |
| "step": 1455, | |
| "valid_targets_mean": 3726.8, | |
| "valid_targets_min": 737 | |
| }, | |
| { | |
| "epoch": 2.6118067978533093, | |
| "grad_norm": 0.6037481930684999, | |
| "learning_rate": 3.160069770181411e-05, | |
| "loss": 0.2875, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3057287335395813, | |
| "step": 1460, | |
| "valid_targets_mean": 3358.2, | |
| "valid_targets_min": 1491 | |
| }, | |
| { | |
| "epoch": 2.620751341681574, | |
| "grad_norm": 0.504749731879137, | |
| "learning_rate": 3.152790101635408e-05, | |
| "loss": 0.2903, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2742614150047302, | |
| "step": 1465, | |
| "valid_targets_mean": 4562.9, | |
| "valid_targets_min": 710 | |
| }, | |
| { | |
| "epoch": 2.629695885509839, | |
| "grad_norm": 0.5501576116598611, | |
| "learning_rate": 3.145487489718692e-05, | |
| "loss": 0.3005, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.29383790493011475, | |
| "step": 1470, | |
| "valid_targets_mean": 4014.8, | |
| "valid_targets_min": 1590 | |
| }, | |
| { | |
| "epoch": 2.638640429338104, | |
| "grad_norm": 0.6055643551418105, | |
| "learning_rate": 3.138162079771278e-05, | |
| "loss": 0.2971, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.30859410762786865, | |
| "step": 1475, | |
| "valid_targets_mean": 3838.2, | |
| "valid_targets_min": 1093 | |
| }, | |
| { | |
| "epoch": 2.6475849731663685, | |
| "grad_norm": 0.6069047010077838, | |
| "learning_rate": 3.1308140175869216e-05, | |
| "loss": 0.2978, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.29912006855010986, | |
| "step": 1480, | |
| "valid_targets_mean": 3958.1, | |
| "valid_targets_min": 1677 | |
| }, | |
| { | |
| "epoch": 2.6565295169946332, | |
| "grad_norm": 0.5738673372037972, | |
| "learning_rate": 3.123443449410211e-05, | |
| "loss": 0.291, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.29722023010253906, | |
| "step": 1485, | |
| "valid_targets_mean": 3427.4, | |
| "valid_targets_min": 527 | |
| }, | |
| { | |
| "epoch": 2.665474060822898, | |
| "grad_norm": 0.5133193729173582, | |
| "learning_rate": 3.1160505219336594e-05, | |
| "loss": 0.3037, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26430267095565796, | |
| "step": 1490, | |
| "valid_targets_mean": 4530.4, | |
| "valid_targets_min": 1263 | |
| }, | |
| { | |
| "epoch": 2.6744186046511627, | |
| "grad_norm": 0.5734809599747189, | |
| "learning_rate": 3.108635382294787e-05, | |
| "loss": 0.2919, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2972019910812378, | |
| "step": 1495, | |
| "valid_targets_mean": 3677.4, | |
| "valid_targets_min": 1602 | |
| }, | |
| { | |
| "epoch": 2.683363148479428, | |
| "grad_norm": 0.5428678152914708, | |
| "learning_rate": 3.101198178073189e-05, | |
| "loss": 0.2907, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2788999080657959, | |
| "step": 1500, | |
| "valid_targets_mean": 4050.4, | |
| "valid_targets_min": 1841 | |
| }, | |
| { | |
| "epoch": 2.6923076923076925, | |
| "grad_norm": 0.5346131939463955, | |
| "learning_rate": 3.093739057287603e-05, | |
| "loss": 0.2943, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3178672790527344, | |
| "step": 1505, | |
| "valid_targets_mean": 4216.1, | |
| "valid_targets_min": 2013 | |
| }, | |
| { | |
| "epoch": 2.701252236135957, | |
| "grad_norm": 0.5546663057017248, | |
| "learning_rate": 3.086258168392957e-05, | |
| "loss": 0.2927, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2941766381263733, | |
| "step": 1510, | |
| "valid_targets_mean": 4037.4, | |
| "valid_targets_min": 2094 | |
| }, | |
| { | |
| "epoch": 2.710196779964222, | |
| "grad_norm": 0.6007631310576768, | |
| "learning_rate": 3.0787556602774195e-05, | |
| "loss": 0.2995, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3104172348976135, | |
| "step": 1515, | |
| "valid_targets_mean": 3226.3, | |
| "valid_targets_min": 699 | |
| }, | |
| { | |
| "epoch": 2.7191413237924866, | |
| "grad_norm": 0.5246652173150378, | |
| "learning_rate": 3.071231682259437e-05, | |
| "loss": 0.2993, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.31006383895874023, | |
| "step": 1520, | |
| "valid_targets_mean": 3979.1, | |
| "valid_targets_min": 1929 | |
| }, | |
| { | |
| "epoch": 2.7280858676207513, | |
| "grad_norm": 0.4917003682982894, | |
| "learning_rate": 3.063686384084756e-05, | |
| "loss": 0.2894, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3010290265083313, | |
| "step": 1525, | |
| "valid_targets_mean": 4885.2, | |
| "valid_targets_min": 1695 | |
| }, | |
| { | |
| "epoch": 2.737030411449016, | |
| "grad_norm": 0.5164120363148196, | |
| "learning_rate": 3.05611991592345e-05, | |
| "loss": 0.292, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28511500358581543, | |
| "step": 1530, | |
| "valid_targets_mean": 4332.8, | |
| "valid_targets_min": 2130 | |
| }, | |
| { | |
| "epoch": 2.7459749552772807, | |
| "grad_norm": 0.546710521171786, | |
| "learning_rate": 3.0485324283669288e-05, | |
| "loss": 0.2976, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2977195382118225, | |
| "step": 1535, | |
| "valid_targets_mean": 3932.6, | |
| "valid_targets_min": 835 | |
| }, | |
| { | |
| "epoch": 2.7549194991055455, | |
| "grad_norm": 0.4987699796804533, | |
| "learning_rate": 3.0409240724249334e-05, | |
| "loss": 0.2974, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28613370656967163, | |
| "step": 1540, | |
| "valid_targets_mean": 4534.4, | |
| "valid_targets_min": 932 | |
| }, | |
| { | |
| "epoch": 2.76386404293381, | |
| "grad_norm": 0.5318282523003093, | |
| "learning_rate": 3.033294999522545e-05, | |
| "loss": 0.291, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.29021310806274414, | |
| "step": 1545, | |
| "valid_targets_mean": 4185.5, | |
| "valid_targets_min": 1805 | |
| }, | |
| { | |
| "epoch": 2.772808586762075, | |
| "grad_norm": 0.5878956920224669, | |
| "learning_rate": 3.0256453614971594e-05, | |
| "loss": 0.3034, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2993810772895813, | |
| "step": 1550, | |
| "valid_targets_mean": 3415.2, | |
| "valid_targets_min": 715 | |
| }, | |
| { | |
| "epoch": 2.78175313059034, | |
| "grad_norm": 0.5178111135961665, | |
| "learning_rate": 3.017975310595469e-05, | |
| "loss": 0.3122, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.32516568899154663, | |
| "step": 1555, | |
| "valid_targets_mean": 4719.7, | |
| "valid_targets_min": 2013 | |
| }, | |
| { | |
| "epoch": 2.7906976744186047, | |
| "grad_norm": 0.5581315425684471, | |
| "learning_rate": 3.0102849994704343e-05, | |
| "loss": 0.3007, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2962647080421448, | |
| "step": 1560, | |
| "valid_targets_mean": 3870.3, | |
| "valid_targets_min": 1774 | |
| }, | |
| { | |
| "epoch": 2.7996422182468694, | |
| "grad_norm": 0.6256471631470097, | |
| "learning_rate": 3.0025745811782444e-05, | |
| "loss": 0.3041, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.29967325925827026, | |
| "step": 1565, | |
| "valid_targets_mean": 3177.1, | |
| "valid_targets_min": 772 | |
| }, | |
| { | |
| "epoch": 2.808586762075134, | |
| "grad_norm": 0.5320691282381641, | |
| "learning_rate": 2.994844209175269e-05, | |
| "loss": 0.2977, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.331190824508667, | |
| "step": 1570, | |
| "valid_targets_mean": 4638.9, | |
| "valid_targets_min": 1626 | |
| }, | |
| { | |
| "epoch": 2.817531305903399, | |
| "grad_norm": 0.6752470008469027, | |
| "learning_rate": 2.987094037315008e-05, | |
| "loss": 0.2964, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3196222186088562, | |
| "step": 1575, | |
| "valid_targets_mean": 2458.3, | |
| "valid_targets_min": 658 | |
| }, | |
| { | |
| "epoch": 2.8264758497316635, | |
| "grad_norm": 0.5531523191801033, | |
| "learning_rate": 2.9793242198450258e-05, | |
| "loss": 0.2781, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28398263454437256, | |
| "step": 1580, | |
| "valid_targets_mean": 4128.9, | |
| "valid_targets_min": 1775 | |
| }, | |
| { | |
| "epoch": 2.8354203935599287, | |
| "grad_norm": 0.5237471309819827, | |
| "learning_rate": 2.9715349114038825e-05, | |
| "loss": 0.2842, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2685989737510681, | |
| "step": 1585, | |
| "valid_targets_mean": 5779.9, | |
| "valid_targets_min": 1305 | |
| }, | |
| { | |
| "epoch": 2.8443649373881934, | |
| "grad_norm": 0.47372842550809013, | |
| "learning_rate": 2.9637262670180597e-05, | |
| "loss": 0.2924, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.29568806290626526, | |
| "step": 1590, | |
| "valid_targets_mean": 4608.0, | |
| "valid_targets_min": 607 | |
| }, | |
| { | |
| "epoch": 2.853309481216458, | |
| "grad_norm": 0.5192555722225825, | |
| "learning_rate": 2.955898442098869e-05, | |
| "loss": 0.2932, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28541460633277893, | |
| "step": 1595, | |
| "valid_targets_mean": 4063.2, | |
| "valid_targets_min": 919 | |
| }, | |
| { | |
| "epoch": 2.862254025044723, | |
| "grad_norm": 0.5309547559574028, | |
| "learning_rate": 2.948051592439363e-05, | |
| "loss": 0.2887, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.31091922521591187, | |
| "step": 1600, | |
| "valid_targets_mean": 3888.9, | |
| "valid_targets_min": 823 | |
| }, | |
| { | |
| "epoch": 2.8711985688729875, | |
| "grad_norm": 0.5461932654893836, | |
| "learning_rate": 2.9401858742112334e-05, | |
| "loss": 0.2952, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.30643460154533386, | |
| "step": 1605, | |
| "valid_targets_mean": 3989.2, | |
| "valid_targets_min": 1666 | |
| }, | |
| { | |
| "epoch": 2.8801431127012522, | |
| "grad_norm": 0.5151199812668427, | |
| "learning_rate": 2.9323014439617044e-05, | |
| "loss": 0.2969, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3191748559474945, | |
| "step": 1610, | |
| "valid_targets_mean": 4541.9, | |
| "valid_targets_min": 2209 | |
| }, | |
| { | |
| "epoch": 2.889087656529517, | |
| "grad_norm": 0.452227709754697, | |
| "learning_rate": 2.924398458610414e-05, | |
| "loss": 0.2677, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23286093771457672, | |
| "step": 1615, | |
| "valid_targets_mean": 4696.9, | |
| "valid_targets_min": 2143 | |
| }, | |
| { | |
| "epoch": 2.8980322003577816, | |
| "grad_norm": 0.5294903977491242, | |
| "learning_rate": 2.9164770754462926e-05, | |
| "loss": 0.2983, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27453750371932983, | |
| "step": 1620, | |
| "valid_targets_mean": 4409.6, | |
| "valid_targets_min": 2448 | |
| }, | |
| { | |
| "epoch": 2.9069767441860463, | |
| "grad_norm": 0.5116227108871184, | |
| "learning_rate": 2.9085374521244333e-05, | |
| "loss": 0.2839, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2921096682548523, | |
| "step": 1625, | |
| "valid_targets_mean": 4203.6, | |
| "valid_targets_min": 774 | |
| }, | |
| { | |
| "epoch": 2.915921288014311, | |
| "grad_norm": 0.8886084421935505, | |
| "learning_rate": 2.900579746662954e-05, | |
| "loss": 0.2994, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2785966992378235, | |
| "step": 1630, | |
| "valid_targets_mean": 4256.9, | |
| "valid_targets_min": 1439 | |
| }, | |
| { | |
| "epoch": 2.9248658318425758, | |
| "grad_norm": 0.5165202216086147, | |
| "learning_rate": 2.8926041174398496e-05, | |
| "loss": 0.2903, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2874126434326172, | |
| "step": 1635, | |
| "valid_targets_mean": 3910.6, | |
| "valid_targets_min": 1567 | |
| }, | |
| { | |
| "epoch": 2.933810375670841, | |
| "grad_norm": 0.5227694759657353, | |
| "learning_rate": 2.8846107231898445e-05, | |
| "loss": 0.2982, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2759662866592407, | |
| "step": 1640, | |
| "valid_targets_mean": 3917.8, | |
| "valid_targets_min": 928 | |
| }, | |
| { | |
| "epoch": 2.9427549194991056, | |
| "grad_norm": 0.610425871678501, | |
| "learning_rate": 2.8765997230012295e-05, | |
| "loss": 0.2952, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28333795070648193, | |
| "step": 1645, | |
| "valid_targets_mean": 4170.1, | |
| "valid_targets_min": 2193 | |
| }, | |
| { | |
| "epoch": 2.9516994633273703, | |
| "grad_norm": 0.46116236306935077, | |
| "learning_rate": 2.868571276312698e-05, | |
| "loss": 0.2866, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28075987100601196, | |
| "step": 1650, | |
| "valid_targets_mean": 4360.5, | |
| "valid_targets_min": 1737 | |
| }, | |
| { | |
| "epoch": 2.960644007155635, | |
| "grad_norm": 0.5095418283669151, | |
| "learning_rate": 2.860525542910171e-05, | |
| "loss": 0.2839, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2718711495399475, | |
| "step": 1655, | |
| "valid_targets_mean": 3928.2, | |
| "valid_targets_min": 1604 | |
| }, | |
| { | |
| "epoch": 2.9695885509838997, | |
| "grad_norm": 0.5085610495904306, | |
| "learning_rate": 2.852462682923619e-05, | |
| "loss": 0.2815, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2917214334011078, | |
| "step": 1660, | |
| "valid_targets_mean": 3901.8, | |
| "valid_targets_min": 1275 | |
| }, | |
| { | |
| "epoch": 2.9785330948121644, | |
| "grad_norm": 0.5658892685832217, | |
| "learning_rate": 2.844382856823872e-05, | |
| "loss": 0.2934, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.29110509157180786, | |
| "step": 1665, | |
| "valid_targets_mean": 3285.5, | |
| "valid_targets_min": 690 | |
| }, | |
| { | |
| "epoch": 2.9874776386404296, | |
| "grad_norm": 0.6573066368722159, | |
| "learning_rate": 2.8362862254194298e-05, | |
| "loss": 0.29, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28502312302589417, | |
| "step": 1670, | |
| "valid_targets_mean": 4089.2, | |
| "valid_targets_min": 787 | |
| }, | |
| { | |
| "epoch": 2.9964221824686943, | |
| "grad_norm": 0.5079392607819926, | |
| "learning_rate": 2.8281729498532574e-05, | |
| "loss": 0.312, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3002755045890808, | |
| "step": 1675, | |
| "valid_targets_mean": 4365.2, | |
| "valid_targets_min": 1049 | |
| }, | |
| { | |
| "epoch": 3.005366726296959, | |
| "grad_norm": 0.5393232205748715, | |
| "learning_rate": 2.8200431915995805e-05, | |
| "loss": 0.2826, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28886353969573975, | |
| "step": 1680, | |
| "valid_targets_mean": 3904.8, | |
| "valid_targets_min": 863 | |
| }, | |
| { | |
| "epoch": 3.0143112701252237, | |
| "grad_norm": 0.5476424824827836, | |
| "learning_rate": 2.8118971124606712e-05, | |
| "loss": 0.2629, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2710658311843872, | |
| "step": 1685, | |
| "valid_targets_mean": 4062.5, | |
| "valid_targets_min": 2017 | |
| }, | |
| { | |
| "epoch": 3.0232558139534884, | |
| "grad_norm": 0.606499437576894, | |
| "learning_rate": 2.8037348745636274e-05, | |
| "loss": 0.2872, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2897275388240814, | |
| "step": 1690, | |
| "valid_targets_mean": 3168.9, | |
| "valid_targets_min": 1034 | |
| }, | |
| { | |
| "epoch": 3.032200357781753, | |
| "grad_norm": 0.5222073176909986, | |
| "learning_rate": 2.7955566403571464e-05, | |
| "loss": 0.2845, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26425305008888245, | |
| "step": 1695, | |
| "valid_targets_mean": 4216.6, | |
| "valid_targets_min": 1906 | |
| }, | |
| { | |
| "epoch": 3.041144901610018, | |
| "grad_norm": 0.533081866038887, | |
| "learning_rate": 2.78736257260829e-05, | |
| "loss": 0.2724, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2789760231971741, | |
| "step": 1700, | |
| "valid_targets_mean": 4355.0, | |
| "valid_targets_min": 1674 | |
| }, | |
| { | |
| "epoch": 3.0500894454382825, | |
| "grad_norm": 0.5395830050252504, | |
| "learning_rate": 2.7791528343992494e-05, | |
| "loss": 0.2914, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2874540090560913, | |
| "step": 1705, | |
| "valid_targets_mean": 3905.8, | |
| "valid_targets_min": 1662 | |
| }, | |
| { | |
| "epoch": 3.0590339892665472, | |
| "grad_norm": 0.5327375583639009, | |
| "learning_rate": 2.7709275891240936e-05, | |
| "loss": 0.2751, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2773253917694092, | |
| "step": 1710, | |
| "valid_targets_mean": 3790.8, | |
| "valid_targets_min": 1228 | |
| }, | |
| { | |
| "epoch": 3.067978533094812, | |
| "grad_norm": 0.5332213202740119, | |
| "learning_rate": 2.7626870004855236e-05, | |
| "loss": 0.2732, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2752811312675476, | |
| "step": 1715, | |
| "valid_targets_mean": 4314.0, | |
| "valid_targets_min": 2290 | |
| }, | |
| { | |
| "epoch": 3.076923076923077, | |
| "grad_norm": 0.5771000275440432, | |
| "learning_rate": 2.7544312324916088e-05, | |
| "loss": 0.2824, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3129065930843353, | |
| "step": 1720, | |
| "valid_targets_mean": 3639.2, | |
| "valid_targets_min": 1703 | |
| }, | |
| { | |
| "epoch": 3.085867620751342, | |
| "grad_norm": 0.4926160303249595, | |
| "learning_rate": 2.7461604494525257e-05, | |
| "loss": 0.2681, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27431702613830566, | |
| "step": 1725, | |
| "valid_targets_mean": 4959.2, | |
| "valid_targets_min": 1500 | |
| }, | |
| { | |
| "epoch": 3.0948121645796065, | |
| "grad_norm": 0.5440762244035418, | |
| "learning_rate": 2.7378748159772888e-05, | |
| "loss": 0.2668, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2572042942047119, | |
| "step": 1730, | |
| "valid_targets_mean": 3845.3, | |
| "valid_targets_min": 867 | |
| }, | |
| { | |
| "epoch": 3.103756708407871, | |
| "grad_norm": 0.5468975063253847, | |
| "learning_rate": 2.7295744969704725e-05, | |
| "loss": 0.2654, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2830193340778351, | |
| "step": 1735, | |
| "valid_targets_mean": 3994.4, | |
| "valid_targets_min": 793 | |
| }, | |
| { | |
| "epoch": 3.112701252236136, | |
| "grad_norm": 0.5247791983633943, | |
| "learning_rate": 2.7212596576289264e-05, | |
| "loss": 0.2749, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25497889518737793, | |
| "step": 1740, | |
| "valid_targets_mean": 3932.6, | |
| "valid_targets_min": 867 | |
| }, | |
| { | |
| "epoch": 3.1216457960644006, | |
| "grad_norm": 0.5097180897051293, | |
| "learning_rate": 2.712930463438496e-05, | |
| "loss": 0.2764, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2678571343421936, | |
| "step": 1745, | |
| "valid_targets_mean": 4393.1, | |
| "valid_targets_min": 1763 | |
| }, | |
| { | |
| "epoch": 3.1305903398926653, | |
| "grad_norm": 0.5054371550866157, | |
| "learning_rate": 2.7045870801707194e-05, | |
| "loss": 0.2761, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2683069109916687, | |
| "step": 1750, | |
| "valid_targets_mean": 4437.9, | |
| "valid_targets_min": 1725 | |
| }, | |
| { | |
| "epoch": 3.13953488372093, | |
| "grad_norm": 0.5389786007076532, | |
| "learning_rate": 2.6962296738795344e-05, | |
| "loss": 0.2906, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.29220160841941833, | |
| "step": 1755, | |
| "valid_targets_mean": 4123.8, | |
| "valid_targets_min": 579 | |
| }, | |
| { | |
| "epoch": 3.148479427549195, | |
| "grad_norm": 0.5762090201207496, | |
| "learning_rate": 2.687858410897971e-05, | |
| "loss": 0.2636, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25318244099617004, | |
| "step": 1760, | |
| "valid_targets_mean": 3570.9, | |
| "valid_targets_min": 743 | |
| }, | |
| { | |
| "epoch": 3.15742397137746, | |
| "grad_norm": 0.5279696697699028, | |
| "learning_rate": 2.679473457834842e-05, | |
| "loss": 0.2916, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.29287105798721313, | |
| "step": 1765, | |
| "valid_targets_mean": 4093.7, | |
| "valid_targets_min": 2342 | |
| }, | |
| { | |
| "epoch": 3.1663685152057246, | |
| "grad_norm": 0.6183882195312416, | |
| "learning_rate": 2.6710749815714262e-05, | |
| "loss": 0.2707, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26889514923095703, | |
| "step": 1770, | |
| "valid_targets_mean": 3194.8, | |
| "valid_targets_min": 676 | |
| }, | |
| { | |
| "epoch": 3.1753130590339893, | |
| "grad_norm": 0.5656736312850078, | |
| "learning_rate": 2.6626631492581475e-05, | |
| "loss": 0.2834, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28966403007507324, | |
| "step": 1775, | |
| "valid_targets_mean": 3782.7, | |
| "valid_targets_min": 1258 | |
| }, | |
| { | |
| "epoch": 3.184257602862254, | |
| "grad_norm": 0.5484222792379775, | |
| "learning_rate": 2.654238128311249e-05, | |
| "loss": 0.2845, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2807530164718628, | |
| "step": 1780, | |
| "valid_targets_mean": 3794.1, | |
| "valid_targets_min": 1944 | |
| }, | |
| { | |
| "epoch": 3.1932021466905187, | |
| "grad_norm": 0.5453822622258856, | |
| "learning_rate": 2.645800086409458e-05, | |
| "loss": 0.2761, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2747189402580261, | |
| "step": 1785, | |
| "valid_targets_mean": 4427.3, | |
| "valid_targets_min": 1608 | |
| }, | |
| { | |
| "epoch": 3.2021466905187834, | |
| "grad_norm": 0.5001051263184161, | |
| "learning_rate": 2.637349191490654e-05, | |
| "loss": 0.2796, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2797410488128662, | |
| "step": 1790, | |
| "valid_targets_mean": 4777.8, | |
| "valid_targets_min": 2640 | |
| }, | |
| { | |
| "epoch": 3.211091234347048, | |
| "grad_norm": 0.5575267617456836, | |
| "learning_rate": 2.6288856117485216e-05, | |
| "loss": 0.2896, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.298971951007843, | |
| "step": 1795, | |
| "valid_targets_mean": 4380.4, | |
| "valid_targets_min": 1213 | |
| }, | |
| { | |
| "epoch": 3.220035778175313, | |
| "grad_norm": 0.5382921298878789, | |
| "learning_rate": 2.6204095156292048e-05, | |
| "loss": 0.2644, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26511919498443604, | |
| "step": 1800, | |
| "valid_targets_mean": 4108.1, | |
| "valid_targets_min": 2044 | |
| }, | |
| { | |
| "epoch": 3.228980322003578, | |
| "grad_norm": 0.5252112327673899, | |
| "learning_rate": 2.6119210718279538e-05, | |
| "loss": 0.2798, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28953590989112854, | |
| "step": 1805, | |
| "valid_targets_mean": 4361.9, | |
| "valid_targets_min": 1299 | |
| }, | |
| { | |
| "epoch": 3.2379248658318427, | |
| "grad_norm": 0.5410954452724649, | |
| "learning_rate": 2.60342044928577e-05, | |
| "loss": 0.2808, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.30843818187713623, | |
| "step": 1810, | |
| "valid_targets_mean": 4546.8, | |
| "valid_targets_min": 1329 | |
| }, | |
| { | |
| "epoch": 3.2468694096601074, | |
| "grad_norm": 0.581674111347238, | |
| "learning_rate": 2.5949078171860395e-05, | |
| "loss": 0.2845, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.30058762431144714, | |
| "step": 1815, | |
| "valid_targets_mean": 3899.8, | |
| "valid_targets_min": 763 | |
| }, | |
| { | |
| "epoch": 3.255813953488372, | |
| "grad_norm": 0.5237279678403152, | |
| "learning_rate": 2.5863833449511706e-05, | |
| "loss": 0.2649, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27064049243927, | |
| "step": 1820, | |
| "valid_targets_mean": 4455.4, | |
| "valid_targets_min": 1818 | |
| }, | |
| { | |
| "epoch": 3.264758497316637, | |
| "grad_norm": 0.5533963196851126, | |
| "learning_rate": 2.5778472022392184e-05, | |
| "loss": 0.2757, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2702397108078003, | |
| "step": 1825, | |
| "valid_targets_mean": 4271.4, | |
| "valid_targets_min": 2002 | |
| }, | |
| { | |
| "epoch": 3.2737030411449015, | |
| "grad_norm": 0.5313362157055852, | |
| "learning_rate": 2.5692995589405087e-05, | |
| "loss": 0.2906, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2803131341934204, | |
| "step": 1830, | |
| "valid_targets_mean": 4208.0, | |
| "valid_targets_min": 1639 | |
| }, | |
| { | |
| "epoch": 3.282647584973166, | |
| "grad_norm": 0.5541192810579373, | |
| "learning_rate": 2.5607405851742578e-05, | |
| "loss": 0.2786, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2655271291732788, | |
| "step": 1835, | |
| "valid_targets_mean": 4423.6, | |
| "valid_targets_min": 1631 | |
| }, | |
| { | |
| "epoch": 3.2915921288014314, | |
| "grad_norm": 0.581829811898252, | |
| "learning_rate": 2.5521704512851884e-05, | |
| "loss": 0.2836, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28082770109176636, | |
| "step": 1840, | |
| "valid_targets_mean": 3964.4, | |
| "valid_targets_min": 1481 | |
| }, | |
| { | |
| "epoch": 3.300536672629696, | |
| "grad_norm": 0.5564304499542477, | |
| "learning_rate": 2.5435893278401328e-05, | |
| "loss": 0.2768, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2610231637954712, | |
| "step": 1845, | |
| "valid_targets_mean": 3748.6, | |
| "valid_targets_min": 1052 | |
| }, | |
| { | |
| "epoch": 3.309481216457961, | |
| "grad_norm": 0.5890948907758241, | |
| "learning_rate": 2.534997385624647e-05, | |
| "loss": 0.2651, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2398790717124939, | |
| "step": 1850, | |
| "valid_targets_mean": 3603.0, | |
| "valid_targets_min": 552 | |
| }, | |
| { | |
| "epoch": 3.3184257602862255, | |
| "grad_norm": 0.563595381422207, | |
| "learning_rate": 2.5263947956396043e-05, | |
| "loss": 0.2768, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2689689099788666, | |
| "step": 1855, | |
| "valid_targets_mean": 3884.8, | |
| "valid_targets_min": 922 | |
| }, | |
| { | |
| "epoch": 3.32737030411449, | |
| "grad_norm": 0.5202754657752621, | |
| "learning_rate": 2.5177817290977967e-05, | |
| "loss": 0.275, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2853309214115143, | |
| "step": 1860, | |
| "valid_targets_mean": 4258.4, | |
| "valid_targets_min": 1557 | |
| }, | |
| { | |
| "epoch": 3.336314847942755, | |
| "grad_norm": 0.5125955462826816, | |
| "learning_rate": 2.5091583574205247e-05, | |
| "loss": 0.2813, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26263585686683655, | |
| "step": 1865, | |
| "valid_targets_mean": 4235.4, | |
| "valid_targets_min": 2097 | |
| }, | |
| { | |
| "epoch": 3.3452593917710196, | |
| "grad_norm": 0.5366296714092318, | |
| "learning_rate": 2.5005248522341868e-05, | |
| "loss": 0.2741, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2572114169597626, | |
| "step": 1870, | |
| "valid_targets_mean": 3658.3, | |
| "valid_targets_min": 1843 | |
| }, | |
| { | |
| "epoch": 3.3542039355992843, | |
| "grad_norm": 0.5473866663101812, | |
| "learning_rate": 2.4918813853668632e-05, | |
| "loss": 0.2807, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2796470522880554, | |
| "step": 1875, | |
| "valid_targets_mean": 3708.9, | |
| "valid_targets_min": 1482 | |
| }, | |
| { | |
| "epoch": 3.363148479427549, | |
| "grad_norm": 0.5566454634749927, | |
| "learning_rate": 2.483228128844896e-05, | |
| "loss": 0.2715, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2780247926712036, | |
| "step": 1880, | |
| "valid_targets_mean": 3811.8, | |
| "valid_targets_min": 842 | |
| }, | |
| { | |
| "epoch": 3.3720930232558137, | |
| "grad_norm": 0.5573845593232436, | |
| "learning_rate": 2.4745652548894654e-05, | |
| "loss": 0.273, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2704911231994629, | |
| "step": 1885, | |
| "valid_targets_mean": 4766.7, | |
| "valid_targets_min": 2080 | |
| }, | |
| { | |
| "epoch": 3.381037567084079, | |
| "grad_norm": 0.5104062442322098, | |
| "learning_rate": 2.4658929359131634e-05, | |
| "loss": 0.273, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2546011209487915, | |
| "step": 1890, | |
| "valid_targets_mean": 4313.2, | |
| "valid_targets_min": 993 | |
| }, | |
| { | |
| "epoch": 3.3899821109123436, | |
| "grad_norm": 0.6120011359863383, | |
| "learning_rate": 2.4572113445165603e-05, | |
| "loss": 0.2944, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.29367899894714355, | |
| "step": 1895, | |
| "valid_targets_mean": 3432.7, | |
| "valid_targets_min": 758 | |
| }, | |
| { | |
| "epoch": 3.3989266547406083, | |
| "grad_norm": 0.5470654579653317, | |
| "learning_rate": 2.4485206534847706e-05, | |
| "loss": 0.2786, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2764478921890259, | |
| "step": 1900, | |
| "valid_targets_mean": 3888.6, | |
| "valid_targets_min": 1597 | |
| }, | |
| { | |
| "epoch": 3.407871198568873, | |
| "grad_norm": 0.5631697298734966, | |
| "learning_rate": 2.439821035784014e-05, | |
| "loss": 0.2801, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.29474952816963196, | |
| "step": 1905, | |
| "valid_targets_mean": 3846.1, | |
| "valid_targets_min": 1665 | |
| }, | |
| { | |
| "epoch": 3.4168157423971377, | |
| "grad_norm": 0.522999992256365, | |
| "learning_rate": 2.431112664558173e-05, | |
| "loss": 0.2704, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2620839476585388, | |
| "step": 1910, | |
| "valid_targets_mean": 3582.1, | |
| "valid_targets_min": 1304 | |
| }, | |
| { | |
| "epoch": 3.4257602862254024, | |
| "grad_norm": 0.49262352099697787, | |
| "learning_rate": 2.4223957131253467e-05, | |
| "loss": 0.2743, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26315534114837646, | |
| "step": 1915, | |
| "valid_targets_mean": 4538.1, | |
| "valid_targets_min": 2028 | |
| }, | |
| { | |
| "epoch": 3.434704830053667, | |
| "grad_norm": 0.5515685545090246, | |
| "learning_rate": 2.4136703549744028e-05, | |
| "loss": 0.2888, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2935299873352051, | |
| "step": 1920, | |
| "valid_targets_mean": 4080.0, | |
| "valid_targets_min": 794 | |
| }, | |
| { | |
| "epoch": 3.4436493738819323, | |
| "grad_norm": 0.48476650247349296, | |
| "learning_rate": 2.4049367637615215e-05, | |
| "loss": 0.2964, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2544916272163391, | |
| "step": 1925, | |
| "valid_targets_mean": 4508.7, | |
| "valid_targets_min": 1366 | |
| }, | |
| { | |
| "epoch": 3.452593917710197, | |
| "grad_norm": 0.4860033907502521, | |
| "learning_rate": 2.396195113306742e-05, | |
| "loss": 0.2764, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2550594210624695, | |
| "step": 1930, | |
| "valid_targets_mean": 4433.0, | |
| "valid_targets_min": 1985 | |
| }, | |
| { | |
| "epoch": 3.4615384615384617, | |
| "grad_norm": 0.5353229674635286, | |
| "learning_rate": 2.3874455775905036e-05, | |
| "loss": 0.2666, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2665395736694336, | |
| "step": 1935, | |
| "valid_targets_mean": 3745.3, | |
| "valid_targets_min": 937 | |
| }, | |
| { | |
| "epoch": 3.4704830053667264, | |
| "grad_norm": 0.5901472498319702, | |
| "learning_rate": 2.3786883307501794e-05, | |
| "loss": 0.2822, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.29798996448516846, | |
| "step": 1940, | |
| "valid_targets_mean": 4030.6, | |
| "valid_targets_min": 1812 | |
| }, | |
| { | |
| "epoch": 3.479427549194991, | |
| "grad_norm": 0.5296365133279529, | |
| "learning_rate": 2.3699235470766147e-05, | |
| "loss": 0.2778, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2898460030555725, | |
| "step": 1945, | |
| "valid_targets_mean": 4661.3, | |
| "valid_targets_min": 2092 | |
| }, | |
| { | |
| "epoch": 3.488372093023256, | |
| "grad_norm": 0.6738643909015106, | |
| "learning_rate": 2.3611514010106564e-05, | |
| "loss": 0.2818, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2727169990539551, | |
| "step": 1950, | |
| "valid_targets_mean": 4183.0, | |
| "valid_targets_min": 1207 | |
| }, | |
| { | |
| "epoch": 3.4973166368515205, | |
| "grad_norm": 0.5074455770952165, | |
| "learning_rate": 2.35237206713968e-05, | |
| "loss": 0.265, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24378815293312073, | |
| "step": 1955, | |
| "valid_targets_mean": 4246.9, | |
| "valid_targets_min": 1394 | |
| }, | |
| { | |
| "epoch": 3.506261180679785, | |
| "grad_norm": 0.5374442471219851, | |
| "learning_rate": 2.3435857201941172e-05, | |
| "loss": 0.2717, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27619755268096924, | |
| "step": 1960, | |
| "valid_targets_mean": 3896.1, | |
| "valid_targets_min": 1301 | |
| }, | |
| { | |
| "epoch": 3.51520572450805, | |
| "grad_norm": 0.5087843696552686, | |
| "learning_rate": 2.3347925350439765e-05, | |
| "loss": 0.2799, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2813979387283325, | |
| "step": 1965, | |
| "valid_targets_mean": 3997.4, | |
| "valid_targets_min": 1369 | |
| }, | |
| { | |
| "epoch": 3.5241502683363146, | |
| "grad_norm": 0.5980560849762698, | |
| "learning_rate": 2.3259926866953636e-05, | |
| "loss": 0.2801, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28663375973701477, | |
| "step": 1970, | |
| "valid_targets_mean": 3297.6, | |
| "valid_targets_min": 1305 | |
| }, | |
| { | |
| "epoch": 3.5330948121645798, | |
| "grad_norm": 0.5193924325842352, | |
| "learning_rate": 2.3171863502869993e-05, | |
| "loss": 0.2671, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26521211862564087, | |
| "step": 1975, | |
| "valid_targets_mean": 4117.4, | |
| "valid_targets_min": 2056 | |
| }, | |
| { | |
| "epoch": 3.5420393559928445, | |
| "grad_norm": 0.5449182986186685, | |
| "learning_rate": 2.3083737010867316e-05, | |
| "loss": 0.2765, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26807746291160583, | |
| "step": 1980, | |
| "valid_targets_mean": 3758.1, | |
| "valid_targets_min": 1563 | |
| }, | |
| { | |
| "epoch": 3.550983899821109, | |
| "grad_norm": 0.6191157830270809, | |
| "learning_rate": 2.2995549144880487e-05, | |
| "loss": 0.2805, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.31274452805519104, | |
| "step": 1985, | |
| "valid_targets_mean": 3152.6, | |
| "valid_targets_min": 710 | |
| }, | |
| { | |
| "epoch": 3.559928443649374, | |
| "grad_norm": 0.4986923272606033, | |
| "learning_rate": 2.290730166006589e-05, | |
| "loss": 0.2733, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27044352889060974, | |
| "step": 1990, | |
| "valid_targets_mean": 4526.8, | |
| "valid_targets_min": 1712 | |
| }, | |
| { | |
| "epoch": 3.5688729874776386, | |
| "grad_norm": 0.5291098132220836, | |
| "learning_rate": 2.2818996312766474e-05, | |
| "loss": 0.2784, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28628769516944885, | |
| "step": 1995, | |
| "valid_targets_mean": 4237.6, | |
| "valid_targets_min": 1300 | |
| }, | |
| { | |
| "epoch": 3.5778175313059033, | |
| "grad_norm": 0.5395780033336437, | |
| "learning_rate": 2.2730634860476778e-05, | |
| "loss": 0.2837, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2602778673171997, | |
| "step": 2000, | |
| "valid_targets_mean": 3754.7, | |
| "valid_targets_min": 2238 | |
| }, | |
| { | |
| "epoch": 3.586762075134168, | |
| "grad_norm": 0.5654796524167398, | |
| "learning_rate": 2.264221906180798e-05, | |
| "loss": 0.2853, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26922616362571716, | |
| "step": 2005, | |
| "valid_targets_mean": 3893.6, | |
| "valid_targets_min": 1565 | |
| }, | |
| { | |
| "epoch": 3.595706618962433, | |
| "grad_norm": 0.5747944159759851, | |
| "learning_rate": 2.255375067645289e-05, | |
| "loss": 0.2888, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.29051297903060913, | |
| "step": 2010, | |
| "valid_targets_mean": 3719.8, | |
| "valid_targets_min": 1764 | |
| }, | |
| { | |
| "epoch": 3.604651162790698, | |
| "grad_norm": 1.076215188277447, | |
| "learning_rate": 2.2465231465150902e-05, | |
| "loss": 0.2711, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2821960151195526, | |
| "step": 2015, | |
| "valid_targets_mean": 4530.9, | |
| "valid_targets_min": 1666 | |
| }, | |
| { | |
| "epoch": 3.6135957066189626, | |
| "grad_norm": 0.5041678901004969, | |
| "learning_rate": 2.2376663189653002e-05, | |
| "loss": 0.2773, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25614356994628906, | |
| "step": 2020, | |
| "valid_targets_mean": 4557.1, | |
| "valid_targets_min": 2257 | |
| }, | |
| { | |
| "epoch": 3.6225402504472273, | |
| "grad_norm": 0.49697468351907303, | |
| "learning_rate": 2.2288047612686655e-05, | |
| "loss": 0.2639, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2509356737136841, | |
| "step": 2025, | |
| "valid_targets_mean": 4656.5, | |
| "valid_targets_min": 2178 | |
| }, | |
| { | |
| "epoch": 3.631484794275492, | |
| "grad_norm": 0.5593780074340731, | |
| "learning_rate": 2.219938649792074e-05, | |
| "loss": 0.2768, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2787664532661438, | |
| "step": 2030, | |
| "valid_targets_mean": 3585.8, | |
| "valid_targets_min": 725 | |
| }, | |
| { | |
| "epoch": 3.6404293381037567, | |
| "grad_norm": 0.6058445555907487, | |
| "learning_rate": 2.2110681609930458e-05, | |
| "loss": 0.283, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28542840480804443, | |
| "step": 2035, | |
| "valid_targets_mean": 3754.0, | |
| "valid_targets_min": 1893 | |
| }, | |
| { | |
| "epoch": 3.6493738819320214, | |
| "grad_norm": 0.6592434752132575, | |
| "learning_rate": 2.2021934714162212e-05, | |
| "loss": 0.2771, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.30062779784202576, | |
| "step": 2040, | |
| "valid_targets_mean": 3522.3, | |
| "valid_targets_min": 527 | |
| }, | |
| { | |
| "epoch": 3.658318425760286, | |
| "grad_norm": 0.511485872021097, | |
| "learning_rate": 2.1933147576898447e-05, | |
| "loss": 0.2728, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25937914848327637, | |
| "step": 2045, | |
| "valid_targets_mean": 3873.6, | |
| "valid_targets_min": 1491 | |
| }, | |
| { | |
| "epoch": 3.667262969588551, | |
| "grad_norm": 0.5594642686607454, | |
| "learning_rate": 2.1844321965222525e-05, | |
| "loss": 0.2825, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3036922216415405, | |
| "step": 2050, | |
| "valid_targets_mean": 3970.2, | |
| "valid_targets_min": 1673 | |
| }, | |
| { | |
| "epoch": 3.6762075134168155, | |
| "grad_norm": 0.5162375365075016, | |
| "learning_rate": 2.1755459646983545e-05, | |
| "loss": 0.2824, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27681976556777954, | |
| "step": 2055, | |
| "valid_targets_mean": 4571.4, | |
| "valid_targets_min": 1171 | |
| }, | |
| { | |
| "epoch": 3.6851520572450807, | |
| "grad_norm": 0.507537692969965, | |
| "learning_rate": 2.1666562390761147e-05, | |
| "loss": 0.2799, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2693493664264679, | |
| "step": 2060, | |
| "valid_targets_mean": 5078.4, | |
| "valid_targets_min": 1264 | |
| }, | |
| { | |
| "epoch": 3.6940966010733454, | |
| "grad_norm": 0.5780198663275775, | |
| "learning_rate": 2.1577631965830324e-05, | |
| "loss": 0.2734, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2515581250190735, | |
| "step": 2065, | |
| "valid_targets_mean": 3473.2, | |
| "valid_targets_min": 1525 | |
| }, | |
| { | |
| "epoch": 3.70304114490161, | |
| "grad_norm": 0.5015533752033411, | |
| "learning_rate": 2.1488670142126234e-05, | |
| "loss": 0.2659, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2759404182434082, | |
| "step": 2070, | |
| "valid_targets_mean": 4662.1, | |
| "valid_targets_min": 1440 | |
| }, | |
| { | |
| "epoch": 3.7119856887298748, | |
| "grad_norm": 0.518248052496532, | |
| "learning_rate": 2.13996786902089e-05, | |
| "loss": 0.274, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27556905150413513, | |
| "step": 2075, | |
| "valid_targets_mean": 4112.4, | |
| "valid_targets_min": 2299 | |
| }, | |
| { | |
| "epoch": 3.7209302325581395, | |
| "grad_norm": 0.53651912472871, | |
| "learning_rate": 2.1310659381228066e-05, | |
| "loss": 0.2846, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2784693241119385, | |
| "step": 2080, | |
| "valid_targets_mean": 3839.6, | |
| "valid_targets_min": 1588 | |
| }, | |
| { | |
| "epoch": 3.729874776386404, | |
| "grad_norm": 0.5764048848203882, | |
| "learning_rate": 2.122161398688788e-05, | |
| "loss": 0.2903, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28273123502731323, | |
| "step": 2085, | |
| "valid_targets_mean": 3669.2, | |
| "valid_targets_min": 741 | |
| }, | |
| { | |
| "epoch": 3.738819320214669, | |
| "grad_norm": 0.4820289047101863, | |
| "learning_rate": 2.1132544279411655e-05, | |
| "loss": 0.2747, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26568305492401123, | |
| "step": 2090, | |
| "valid_targets_mean": 4602.2, | |
| "valid_targets_min": 786 | |
| }, | |
| { | |
| "epoch": 3.747763864042934, | |
| "grad_norm": 0.6120891450370751, | |
| "learning_rate": 2.1043452031506578e-05, | |
| "loss": 0.2807, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2671006917953491, | |
| "step": 2095, | |
| "valid_targets_mean": 3208.0, | |
| "valid_targets_min": 1621 | |
| }, | |
| { | |
| "epoch": 3.7567084078711988, | |
| "grad_norm": 0.4891187435236056, | |
| "learning_rate": 2.0954339016328485e-05, | |
| "loss": 0.2781, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25922301411628723, | |
| "step": 2100, | |
| "valid_targets_mean": 4271.0, | |
| "valid_targets_min": 1612 | |
| }, | |
| { | |
| "epoch": 3.7656529516994635, | |
| "grad_norm": 0.5464092724861203, | |
| "learning_rate": 2.08652070074465e-05, | |
| "loss": 0.2805, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3047419786453247, | |
| "step": 2105, | |
| "valid_targets_mean": 3791.9, | |
| "valid_targets_min": 1102 | |
| }, | |
| { | |
| "epoch": 3.774597495527728, | |
| "grad_norm": 0.6364819542700682, | |
| "learning_rate": 2.077605777880778e-05, | |
| "loss": 0.2681, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.31326138973236084, | |
| "step": 2110, | |
| "valid_targets_mean": 3016.8, | |
| "valid_targets_min": 748 | |
| }, | |
| { | |
| "epoch": 3.783542039355993, | |
| "grad_norm": 0.5001257137373921, | |
| "learning_rate": 2.0686893104702198e-05, | |
| "loss": 0.2589, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2531580328941345, | |
| "step": 2115, | |
| "valid_targets_mean": 4102.5, | |
| "valid_targets_min": 1660 | |
| }, | |
| { | |
| "epoch": 3.7924865831842576, | |
| "grad_norm": 0.474779949675362, | |
| "learning_rate": 2.0597714759727028e-05, | |
| "loss": 0.2834, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2678177058696747, | |
| "step": 2120, | |
| "valid_targets_mean": 4861.8, | |
| "valid_targets_min": 1696 | |
| }, | |
| { | |
| "epoch": 3.8014311270125223, | |
| "grad_norm": 0.46813381967025147, | |
| "learning_rate": 2.050852451875163e-05, | |
| "loss": 0.2715, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27863186597824097, | |
| "step": 2125, | |
| "valid_targets_mean": 4995.9, | |
| "valid_targets_min": 715 | |
| }, | |
| { | |
| "epoch": 3.810375670840787, | |
| "grad_norm": 0.5646539409857256, | |
| "learning_rate": 2.041932415688212e-05, | |
| "loss": 0.2749, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2734571099281311, | |
| "step": 2130, | |
| "valid_targets_mean": 4401.4, | |
| "valid_targets_min": 1873 | |
| }, | |
| { | |
| "epoch": 3.8193202146690517, | |
| "grad_norm": 0.5405176564877853, | |
| "learning_rate": 2.0330115449426054e-05, | |
| "loss": 0.2792, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.30330121517181396, | |
| "step": 2135, | |
| "valid_targets_mean": 4075.5, | |
| "valid_targets_min": 988 | |
| }, | |
| { | |
| "epoch": 3.8282647584973164, | |
| "grad_norm": 0.5063431046009021, | |
| "learning_rate": 2.0240900171857064e-05, | |
| "loss": 0.2892, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.29637011885643005, | |
| "step": 2140, | |
| "valid_targets_mean": 4556.2, | |
| "valid_targets_min": 1850 | |
| }, | |
| { | |
| "epoch": 3.8372093023255816, | |
| "grad_norm": 0.6148280852194663, | |
| "learning_rate": 2.0151680099779574e-05, | |
| "loss": 0.2795, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2594483494758606, | |
| "step": 2145, | |
| "valid_targets_mean": 3378.6, | |
| "valid_targets_min": 943 | |
| }, | |
| { | |
| "epoch": 3.8461538461538463, | |
| "grad_norm": 0.5460464073665912, | |
| "learning_rate": 2.0062457008893408e-05, | |
| "loss": 0.2775, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25146251916885376, | |
| "step": 2150, | |
| "valid_targets_mean": 3959.3, | |
| "valid_targets_min": 1579 | |
| }, | |
| { | |
| "epoch": 3.855098389982111, | |
| "grad_norm": 0.5034957296053513, | |
| "learning_rate": 1.9973232674958487e-05, | |
| "loss": 0.2698, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25639331340789795, | |
| "step": 2155, | |
| "valid_targets_mean": 4098.7, | |
| "valid_targets_min": 1449 | |
| }, | |
| { | |
| "epoch": 3.8640429338103757, | |
| "grad_norm": 0.566801459570412, | |
| "learning_rate": 1.9884008873759446e-05, | |
| "loss": 0.2751, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.29593968391418457, | |
| "step": 2160, | |
| "valid_targets_mean": 3769.4, | |
| "valid_targets_min": 474 | |
| }, | |
| { | |
| "epoch": 3.8729874776386404, | |
| "grad_norm": 0.5224664015779718, | |
| "learning_rate": 1.979478738107035e-05, | |
| "loss": 0.2669, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2669827938079834, | |
| "step": 2165, | |
| "valid_targets_mean": 4153.2, | |
| "valid_targets_min": 1824 | |
| }, | |
| { | |
| "epoch": 3.881932021466905, | |
| "grad_norm": 0.5346475973128151, | |
| "learning_rate": 1.9705569972619295e-05, | |
| "loss": 0.2837, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28263264894485474, | |
| "step": 2170, | |
| "valid_targets_mean": 3984.6, | |
| "valid_targets_min": 1518 | |
| }, | |
| { | |
| "epoch": 3.89087656529517, | |
| "grad_norm": 0.5482896143276372, | |
| "learning_rate": 1.961635842405311e-05, | |
| "loss": 0.2675, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27248382568359375, | |
| "step": 2175, | |
| "valid_targets_mean": 4330.0, | |
| "valid_targets_min": 1711 | |
| }, | |
| { | |
| "epoch": 3.899821109123435, | |
| "grad_norm": 0.48204226172382336, | |
| "learning_rate": 1.9527154510901972e-05, | |
| "loss": 0.2734, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25980862975120544, | |
| "step": 2180, | |
| "valid_targets_mean": 4406.8, | |
| "valid_targets_min": 1855 | |
| }, | |
| { | |
| "epoch": 3.9087656529516996, | |
| "grad_norm": 0.4210041453535078, | |
| "learning_rate": 1.9437960008544126e-05, | |
| "loss": 0.2578, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2111448049545288, | |
| "step": 2185, | |
| "valid_targets_mean": 5192.1, | |
| "valid_targets_min": 2200 | |
| }, | |
| { | |
| "epoch": 3.9177101967799643, | |
| "grad_norm": 0.5637363734690357, | |
| "learning_rate": 1.9348776692170494e-05, | |
| "loss": 0.2836, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2881026566028595, | |
| "step": 2190, | |
| "valid_targets_mean": 3882.8, | |
| "valid_targets_min": 558 | |
| }, | |
| { | |
| "epoch": 3.926654740608229, | |
| "grad_norm": 0.48316017438181486, | |
| "learning_rate": 1.9259606336749374e-05, | |
| "loss": 0.2784, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2716853618621826, | |
| "step": 2195, | |
| "valid_targets_mean": 4683.8, | |
| "valid_targets_min": 1674 | |
| }, | |
| { | |
| "epoch": 3.9355992844364938, | |
| "grad_norm": 0.45317704598781144, | |
| "learning_rate": 1.9170450716991122e-05, | |
| "loss": 0.2615, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.236902117729187, | |
| "step": 2200, | |
| "valid_targets_mean": 4864.1, | |
| "valid_targets_min": 2033 | |
| }, | |
| { | |
| "epoch": 3.9445438282647585, | |
| "grad_norm": 0.56647472453608, | |
| "learning_rate": 1.90813116073128e-05, | |
| "loss": 0.2896, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.30802735686302185, | |
| "step": 2205, | |
| "valid_targets_mean": 3714.6, | |
| "valid_targets_min": 823 | |
| }, | |
| { | |
| "epoch": 3.953488372093023, | |
| "grad_norm": 0.5508322430994914, | |
| "learning_rate": 1.899219078180289e-05, | |
| "loss": 0.2789, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.29120323061943054, | |
| "step": 2210, | |
| "valid_targets_mean": 3726.9, | |
| "valid_targets_min": 1807 | |
| }, | |
| { | |
| "epoch": 3.962432915921288, | |
| "grad_norm": 0.5708136652371602, | |
| "learning_rate": 1.8903090014185984e-05, | |
| "loss": 0.2706, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26580384373664856, | |
| "step": 2215, | |
| "valid_targets_mean": 3621.6, | |
| "valid_targets_min": 1470 | |
| }, | |
| { | |
| "epoch": 3.9713774597495526, | |
| "grad_norm": 0.5942416912075751, | |
| "learning_rate": 1.881401107778744e-05, | |
| "loss": 0.2802, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.30616092681884766, | |
| "step": 2220, | |
| "valid_targets_mean": 3367.2, | |
| "valid_targets_min": 777 | |
| }, | |
| { | |
| "epoch": 3.9803220035778173, | |
| "grad_norm": 0.5144304612582596, | |
| "learning_rate": 1.8724955745498147e-05, | |
| "loss": 0.2689, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2784048318862915, | |
| "step": 2225, | |
| "valid_targets_mean": 4234.6, | |
| "valid_targets_min": 1729 | |
| }, | |
| { | |
| "epoch": 3.9892665474060824, | |
| "grad_norm": 0.5300804516751338, | |
| "learning_rate": 1.863592578973921e-05, | |
| "loss": 0.2769, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27351003885269165, | |
| "step": 2230, | |
| "valid_targets_mean": 3583.5, | |
| "valid_targets_min": 1324 | |
| }, | |
| { | |
| "epoch": 3.998211091234347, | |
| "grad_norm": 0.5329781559722259, | |
| "learning_rate": 1.854692298242667e-05, | |
| "loss": 0.2692, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24892058968544006, | |
| "step": 2235, | |
| "valid_targets_mean": 3651.9, | |
| "valid_targets_min": 1519 | |
| }, | |
| { | |
| "epoch": 4.007155635062611, | |
| "grad_norm": 0.4900604244331859, | |
| "learning_rate": 1.8457949094936265e-05, | |
| "loss": 0.2479, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2408735752105713, | |
| "step": 2240, | |
| "valid_targets_mean": 4293.0, | |
| "valid_targets_min": 1617 | |
| }, | |
| { | |
| "epoch": 4.016100178890877, | |
| "grad_norm": 0.49977219265912204, | |
| "learning_rate": 1.8369005898068136e-05, | |
| "loss": 0.25, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26399219036102295, | |
| "step": 2245, | |
| "valid_targets_mean": 5123.9, | |
| "valid_targets_min": 667 | |
| }, | |
| { | |
| "epoch": 4.025044722719142, | |
| "grad_norm": 0.5502105291911087, | |
| "learning_rate": 1.8280095162011617e-05, | |
| "loss": 0.2608, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28108128905296326, | |
| "step": 2250, | |
| "valid_targets_mean": 4024.1, | |
| "valid_targets_min": 747 | |
| }, | |
| { | |
| "epoch": 4.033989266547406, | |
| "grad_norm": 0.5835364945002972, | |
| "learning_rate": 1.819121865630998e-05, | |
| "loss": 0.2627, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25397148728370667, | |
| "step": 2255, | |
| "valid_targets_mean": 3106.8, | |
| "valid_targets_min": 1424 | |
| }, | |
| { | |
| "epoch": 4.042933810375671, | |
| "grad_norm": 0.5736849669987507, | |
| "learning_rate": 1.8102378149825242e-05, | |
| "loss": 0.2566, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25006911158561707, | |
| "step": 2260, | |
| "valid_targets_mean": 4338.4, | |
| "valid_targets_min": 1212 | |
| }, | |
| { | |
| "epoch": 4.051878354203936, | |
| "grad_norm": 0.578856261130234, | |
| "learning_rate": 1.801357541070294e-05, | |
| "loss": 0.2561, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2498929500579834, | |
| "step": 2265, | |
| "valid_targets_mean": 3711.7, | |
| "valid_targets_min": 1454 | |
| }, | |
| { | |
| "epoch": 4.0608228980322005, | |
| "grad_norm": 0.570026305308228, | |
| "learning_rate": 1.7924812206336956e-05, | |
| "loss": 0.2641, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24680036306381226, | |
| "step": 2270, | |
| "valid_targets_mean": 3502.1, | |
| "valid_targets_min": 766 | |
| }, | |
| { | |
| "epoch": 4.069767441860465, | |
| "grad_norm": 0.5017381451878832, | |
| "learning_rate": 1.7836090303334312e-05, | |
| "loss": 0.2587, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24516400694847107, | |
| "step": 2275, | |
| "valid_targets_mean": 4885.2, | |
| "valid_targets_min": 1056 | |
| }, | |
| { | |
| "epoch": 4.07871198568873, | |
| "grad_norm": 0.6571441759078382, | |
| "learning_rate": 1.7747411467480058e-05, | |
| "loss": 0.2619, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25563162565231323, | |
| "step": 2280, | |
| "valid_targets_mean": 4010.9, | |
| "valid_targets_min": 1636 | |
| }, | |
| { | |
| "epoch": 4.087656529516995, | |
| "grad_norm": 0.5157227921600919, | |
| "learning_rate": 1.7658777463702076e-05, | |
| "loss": 0.2611, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26061367988586426, | |
| "step": 2285, | |
| "valid_targets_mean": 4204.8, | |
| "valid_targets_min": 1436 | |
| }, | |
| { | |
| "epoch": 4.096601073345259, | |
| "grad_norm": 0.5245851983426554, | |
| "learning_rate": 1.7570190056035996e-05, | |
| "loss": 0.2598, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23559322953224182, | |
| "step": 2290, | |
| "valid_targets_mean": 4053.4, | |
| "valid_targets_min": 722 | |
| }, | |
| { | |
| "epoch": 4.105545617173524, | |
| "grad_norm": 0.5917926110353966, | |
| "learning_rate": 1.7481651007590054e-05, | |
| "loss": 0.2706, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25014737248420715, | |
| "step": 2295, | |
| "valid_targets_mean": 3289.2, | |
| "valid_targets_min": 699 | |
| }, | |
| { | |
| "epoch": 4.114490161001789, | |
| "grad_norm": 0.5026025456498621, | |
| "learning_rate": 1.7393162080510045e-05, | |
| "loss": 0.2712, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2548500895500183, | |
| "step": 2300, | |
| "valid_targets_mean": 4602.9, | |
| "valid_targets_min": 1525 | |
| }, | |
| { | |
| "epoch": 4.1234347048300535, | |
| "grad_norm": 0.5164574844544229, | |
| "learning_rate": 1.7304725035944194e-05, | |
| "loss": 0.2549, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2356826364994049, | |
| "step": 2305, | |
| "valid_targets_mean": 4163.1, | |
| "valid_targets_min": 1691 | |
| }, | |
| { | |
| "epoch": 4.132379248658318, | |
| "grad_norm": 0.48824332711112567, | |
| "learning_rate": 1.7216341634008168e-05, | |
| "loss": 0.2493, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25585299730300903, | |
| "step": 2310, | |
| "valid_targets_mean": 5345.3, | |
| "valid_targets_min": 2336 | |
| }, | |
| { | |
| "epoch": 4.141323792486583, | |
| "grad_norm": 0.5908232354996049, | |
| "learning_rate": 1.7128013633749978e-05, | |
| "loss": 0.2713, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26593607664108276, | |
| "step": 2315, | |
| "valid_targets_mean": 4050.5, | |
| "valid_targets_min": 684 | |
| }, | |
| { | |
| "epoch": 4.150268336314848, | |
| "grad_norm": 0.5495386431735231, | |
| "learning_rate": 1.7039742793115025e-05, | |
| "loss": 0.2688, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2615734040737152, | |
| "step": 2320, | |
| "valid_targets_mean": 3975.1, | |
| "valid_targets_min": 2183 | |
| }, | |
| { | |
| "epoch": 4.159212880143112, | |
| "grad_norm": 0.6242915835353913, | |
| "learning_rate": 1.6951530868911097e-05, | |
| "loss": 0.2654, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2717490792274475, | |
| "step": 2325, | |
| "valid_targets_mean": 3084.0, | |
| "valid_targets_min": 737 | |
| }, | |
| { | |
| "epoch": 4.168157423971378, | |
| "grad_norm": 0.5070664833456054, | |
| "learning_rate": 1.6863379616773388e-05, | |
| "loss": 0.2641, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2583721876144409, | |
| "step": 2330, | |
| "valid_targets_mean": 4625.7, | |
| "valid_targets_min": 1817 | |
| }, | |
| { | |
| "epoch": 4.177101967799643, | |
| "grad_norm": 0.5582812233226788, | |
| "learning_rate": 1.6775290791129567e-05, | |
| "loss": 0.2584, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2620850205421448, | |
| "step": 2335, | |
| "valid_targets_mean": 3869.9, | |
| "valid_targets_min": 1570 | |
| }, | |
| { | |
| "epoch": 4.186046511627907, | |
| "grad_norm": 0.584227709585007, | |
| "learning_rate": 1.668726614516488e-05, | |
| "loss": 0.2689, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25571852922439575, | |
| "step": 2340, | |
| "valid_targets_mean": 3331.9, | |
| "valid_targets_min": 1615 | |
| }, | |
| { | |
| "epoch": 4.194991055456172, | |
| "grad_norm": 0.5175133185961949, | |
| "learning_rate": 1.6599307430787216e-05, | |
| "loss": 0.256, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2526220977306366, | |
| "step": 2345, | |
| "valid_targets_mean": 4406.4, | |
| "valid_targets_min": 1662 | |
| }, | |
| { | |
| "epoch": 4.203935599284437, | |
| "grad_norm": 0.49289171535932735, | |
| "learning_rate": 1.6511416398592264e-05, | |
| "loss": 0.2705, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2843324840068817, | |
| "step": 2350, | |
| "valid_targets_mean": 5134.6, | |
| "valid_targets_min": 995 | |
| }, | |
| { | |
| "epoch": 4.212880143112701, | |
| "grad_norm": 0.5356859083954757, | |
| "learning_rate": 1.64235947978287e-05, | |
| "loss": 0.2669, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28424036502838135, | |
| "step": 2355, | |
| "valid_targets_mean": 4424.1, | |
| "valid_targets_min": 1810 | |
| }, | |
| { | |
| "epoch": 4.221824686940966, | |
| "grad_norm": 0.513042991430007, | |
| "learning_rate": 1.6335844376363295e-05, | |
| "loss": 0.2664, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2585211396217346, | |
| "step": 2360, | |
| "valid_targets_mean": 5082.6, | |
| "valid_targets_min": 907 | |
| }, | |
| { | |
| "epoch": 4.230769230769231, | |
| "grad_norm": 0.5695128355206176, | |
| "learning_rate": 1.6248166880646225e-05, | |
| "loss": 0.2557, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2641540467739105, | |
| "step": 2365, | |
| "valid_targets_mean": 3779.7, | |
| "valid_targets_min": 1253 | |
| }, | |
| { | |
| "epoch": 4.2397137745974955, | |
| "grad_norm": 0.6023295149757434, | |
| "learning_rate": 1.6160564055676235e-05, | |
| "loss": 0.259, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2857375144958496, | |
| "step": 2370, | |
| "valid_targets_mean": 2948.9, | |
| "valid_targets_min": 720 | |
| }, | |
| { | |
| "epoch": 4.24865831842576, | |
| "grad_norm": 0.5098827801538092, | |
| "learning_rate": 1.6073037644965938e-05, | |
| "loss": 0.2548, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23553355038166046, | |
| "step": 2375, | |
| "valid_targets_mean": 4315.9, | |
| "valid_targets_min": 1473 | |
| }, | |
| { | |
| "epoch": 4.257602862254025, | |
| "grad_norm": 0.5243491333274759, | |
| "learning_rate": 1.5985589390507117e-05, | |
| "loss": 0.2627, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27872568368911743, | |
| "step": 2380, | |
| "valid_targets_mean": 4803.7, | |
| "valid_targets_min": 842 | |
| }, | |
| { | |
| "epoch": 4.26654740608229, | |
| "grad_norm": 0.5241756061317848, | |
| "learning_rate": 1.589822103273607e-05, | |
| "loss": 0.2633, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2604633867740631, | |
| "step": 2385, | |
| "valid_targets_mean": 3930.9, | |
| "valid_targets_min": 835 | |
| }, | |
| { | |
| "epoch": 4.275491949910554, | |
| "grad_norm": 0.4948005196786464, | |
| "learning_rate": 1.5810934310498922e-05, | |
| "loss": 0.2733, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28036582469940186, | |
| "step": 2390, | |
| "valid_targets_mean": 4740.4, | |
| "valid_targets_min": 1665 | |
| }, | |
| { | |
| "epoch": 4.284436493738819, | |
| "grad_norm": 0.5212209771223362, | |
| "learning_rate": 1.5723730961017082e-05, | |
| "loss": 0.2593, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28980493545532227, | |
| "step": 2395, | |
| "valid_targets_mean": 4507.4, | |
| "valid_targets_min": 1489 | |
| }, | |
| { | |
| "epoch": 4.293381037567084, | |
| "grad_norm": 0.5309847222961126, | |
| "learning_rate": 1.5636612719852615e-05, | |
| "loss": 0.2732, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2707955837249756, | |
| "step": 2400, | |
| "valid_targets_mean": 4071.7, | |
| "valid_targets_min": 2048 | |
| }, | |
| { | |
| "epoch": 4.3023255813953485, | |
| "grad_norm": 0.6436096499085671, | |
| "learning_rate": 1.5549581320873715e-05, | |
| "loss": 0.2596, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27491945028305054, | |
| "step": 2405, | |
| "valid_targets_mean": 3519.8, | |
| "valid_targets_min": 483 | |
| }, | |
| { | |
| "epoch": 4.311270125223613, | |
| "grad_norm": 0.6837292335134912, | |
| "learning_rate": 1.5462638496220223e-05, | |
| "loss": 0.2615, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2572469413280487, | |
| "step": 2410, | |
| "valid_targets_mean": 3385.1, | |
| "valid_targets_min": 787 | |
| }, | |
| { | |
| "epoch": 4.320214669051879, | |
| "grad_norm": 0.5148506938846231, | |
| "learning_rate": 1.537578597626912e-05, | |
| "loss": 0.2701, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2704946994781494, | |
| "step": 2415, | |
| "valid_targets_mean": 4277.2, | |
| "valid_targets_min": 1140 | |
| }, | |
| { | |
| "epoch": 4.3291592128801435, | |
| "grad_norm": 0.5488881906738768, | |
| "learning_rate": 1.528902548960009e-05, | |
| "loss": 0.2697, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26252859830856323, | |
| "step": 2420, | |
| "valid_targets_mean": 4134.1, | |
| "valid_targets_min": 697 | |
| }, | |
| { | |
| "epoch": 4.338103756708408, | |
| "grad_norm": 0.728874058237292, | |
| "learning_rate": 1.5202358762961155e-05, | |
| "loss": 0.2694, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2942860424518585, | |
| "step": 2425, | |
| "valid_targets_mean": 2536.9, | |
| "valid_targets_min": 629 | |
| }, | |
| { | |
| "epoch": 4.347048300536673, | |
| "grad_norm": 0.542241600208826, | |
| "learning_rate": 1.511578752123426e-05, | |
| "loss": 0.2574, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2418219894170761, | |
| "step": 2430, | |
| "valid_targets_mean": 3832.9, | |
| "valid_targets_min": 1764 | |
| }, | |
| { | |
| "epoch": 4.355992844364938, | |
| "grad_norm": 0.5211187289766209, | |
| "learning_rate": 1.5029313487400958e-05, | |
| "loss": 0.2681, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27209898829460144, | |
| "step": 2435, | |
| "valid_targets_mean": 4450.8, | |
| "valid_targets_min": 1600 | |
| }, | |
| { | |
| "epoch": 4.364937388193202, | |
| "grad_norm": 0.519412156562896, | |
| "learning_rate": 1.4942938382508147e-05, | |
| "loss": 0.2535, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2514136731624603, | |
| "step": 2440, | |
| "valid_targets_mean": 3935.2, | |
| "valid_targets_min": 1225 | |
| }, | |
| { | |
| "epoch": 4.373881932021467, | |
| "grad_norm": 0.5552648581770736, | |
| "learning_rate": 1.4856663925633776e-05, | |
| "loss": 0.2623, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27997347712516785, | |
| "step": 2445, | |
| "valid_targets_mean": 3901.2, | |
| "valid_targets_min": 656 | |
| }, | |
| { | |
| "epoch": 4.382826475849732, | |
| "grad_norm": 0.5765361718445776, | |
| "learning_rate": 1.4770491833852663e-05, | |
| "loss": 0.2663, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25308501720428467, | |
| "step": 2450, | |
| "valid_targets_mean": 3612.2, | |
| "valid_targets_min": 1505 | |
| }, | |
| { | |
| "epoch": 4.391771019677996, | |
| "grad_norm": 0.4583081364190965, | |
| "learning_rate": 1.4684423822202297e-05, | |
| "loss": 0.2566, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2737816870212555, | |
| "step": 2455, | |
| "valid_targets_mean": 5324.3, | |
| "valid_targets_min": 1435 | |
| }, | |
| { | |
| "epoch": 4.400715563506261, | |
| "grad_norm": 0.5330319752947505, | |
| "learning_rate": 1.4598461603648736e-05, | |
| "loss": 0.2692, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2785816788673401, | |
| "step": 2460, | |
| "valid_targets_mean": 3858.1, | |
| "valid_targets_min": 646 | |
| }, | |
| { | |
| "epoch": 4.409660107334526, | |
| "grad_norm": 0.5365942670180822, | |
| "learning_rate": 1.451260688905245e-05, | |
| "loss": 0.266, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25116056203842163, | |
| "step": 2465, | |
| "valid_targets_mean": 3929.6, | |
| "valid_targets_min": 1495 | |
| }, | |
| { | |
| "epoch": 4.4186046511627906, | |
| "grad_norm": 0.5412560763819989, | |
| "learning_rate": 1.4426861387134361e-05, | |
| "loss": 0.2635, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2546349763870239, | |
| "step": 2470, | |
| "valid_targets_mean": 3853.9, | |
| "valid_targets_min": 1665 | |
| }, | |
| { | |
| "epoch": 4.427549194991055, | |
| "grad_norm": 0.49146455305384606, | |
| "learning_rate": 1.4341226804441766e-05, | |
| "loss": 0.2707, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.284711092710495, | |
| "step": 2475, | |
| "valid_targets_mean": 4519.9, | |
| "valid_targets_min": 2096 | |
| }, | |
| { | |
| "epoch": 4.43649373881932, | |
| "grad_norm": 0.5417709539865344, | |
| "learning_rate": 1.4255704845314406e-05, | |
| "loss": 0.2744, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28545889258384705, | |
| "step": 2480, | |
| "valid_targets_mean": 4169.8, | |
| "valid_targets_min": 1480 | |
| }, | |
| { | |
| "epoch": 4.445438282647585, | |
| "grad_norm": 0.580103049281134, | |
| "learning_rate": 1.4170297211850522e-05, | |
| "loss": 0.2597, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2692874073982239, | |
| "step": 2485, | |
| "valid_targets_mean": 3508.5, | |
| "valid_targets_min": 1796 | |
| }, | |
| { | |
| "epoch": 4.454382826475849, | |
| "grad_norm": 0.518242596266636, | |
| "learning_rate": 1.4085005603873004e-05, | |
| "loss": 0.2748, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27401190996170044, | |
| "step": 2490, | |
| "valid_targets_mean": 4321.2, | |
| "valid_targets_min": 1504 | |
| }, | |
| { | |
| "epoch": 4.463327370304114, | |
| "grad_norm": 0.5757017697078726, | |
| "learning_rate": 1.3999831718895545e-05, | |
| "loss": 0.276, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28794726729393005, | |
| "step": 2495, | |
| "valid_targets_mean": 4045.3, | |
| "valid_targets_min": 1123 | |
| }, | |
| { | |
| "epoch": 4.47227191413238, | |
| "grad_norm": 0.5507231462190035, | |
| "learning_rate": 1.3914777252088851e-05, | |
| "loss": 0.2666, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28651851415634155, | |
| "step": 2500, | |
| "valid_targets_mean": 4500.8, | |
| "valid_targets_min": 1340 | |
| }, | |
| { | |
| "epoch": 4.481216457960644, | |
| "grad_norm": 0.5709744504493519, | |
| "learning_rate": 1.3829843896246917e-05, | |
| "loss": 0.2715, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2755128741264343, | |
| "step": 2505, | |
| "valid_targets_mean": 3492.8, | |
| "valid_targets_min": 1623 | |
| }, | |
| { | |
| "epoch": 4.490161001788909, | |
| "grad_norm": 0.6042383309260438, | |
| "learning_rate": 1.3745033341753337e-05, | |
| "loss": 0.2687, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2556525468826294, | |
| "step": 2510, | |
| "valid_targets_mean": 3758.3, | |
| "valid_targets_min": 1447 | |
| }, | |
| { | |
| "epoch": 4.499105545617174, | |
| "grad_norm": 0.5391378624752455, | |
| "learning_rate": 1.3660347276547645e-05, | |
| "loss": 0.2761, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2629320025444031, | |
| "step": 2515, | |
| "valid_targets_mean": 3977.8, | |
| "valid_targets_min": 1561 | |
| }, | |
| { | |
| "epoch": 4.5080500894454385, | |
| "grad_norm": 0.5894869456529992, | |
| "learning_rate": 1.3575787386091745e-05, | |
| "loss": 0.2549, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2664399743080139, | |
| "step": 2520, | |
| "valid_targets_mean": 3450.4, | |
| "valid_targets_min": 1805 | |
| }, | |
| { | |
| "epoch": 4.516994633273703, | |
| "grad_norm": 0.5927649885745278, | |
| "learning_rate": 1.3491355353336338e-05, | |
| "loss": 0.2701, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2676470875740051, | |
| "step": 2525, | |
| "valid_targets_mean": 3550.7, | |
| "valid_targets_min": 741 | |
| }, | |
| { | |
| "epoch": 4.525939177101968, | |
| "grad_norm": 0.5460664988494238, | |
| "learning_rate": 1.3407052858687432e-05, | |
| "loss": 0.2693, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28948739171028137, | |
| "step": 2530, | |
| "valid_targets_mean": 4106.0, | |
| "valid_targets_min": 1399 | |
| }, | |
| { | |
| "epoch": 4.534883720930233, | |
| "grad_norm": 0.547063854344203, | |
| "learning_rate": 1.3322881579972934e-05, | |
| "loss": 0.2616, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2661360502243042, | |
| "step": 2535, | |
| "valid_targets_mean": 3879.1, | |
| "valid_targets_min": 754 | |
| }, | |
| { | |
| "epoch": 4.543828264758497, | |
| "grad_norm": 0.5551257878105804, | |
| "learning_rate": 1.323884319240921e-05, | |
| "loss": 0.2649, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.30140677094459534, | |
| "step": 2540, | |
| "valid_targets_mean": 3945.9, | |
| "valid_targets_min": 1303 | |
| }, | |
| { | |
| "epoch": 4.552772808586762, | |
| "grad_norm": 0.5712690563559092, | |
| "learning_rate": 1.3154939368567771e-05, | |
| "loss": 0.2432, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25856685638427734, | |
| "step": 2545, | |
| "valid_targets_mean": 3441.0, | |
| "valid_targets_min": 1512 | |
| }, | |
| { | |
| "epoch": 4.561717352415027, | |
| "grad_norm": 0.5434270961027026, | |
| "learning_rate": 1.3071171778341975e-05, | |
| "loss": 0.2597, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25168734788894653, | |
| "step": 2550, | |
| "valid_targets_mean": 4073.9, | |
| "valid_targets_min": 1313 | |
| }, | |
| { | |
| "epoch": 4.5706618962432914, | |
| "grad_norm": 0.5497424956791681, | |
| "learning_rate": 1.2987542088913801e-05, | |
| "loss": 0.2587, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24401748180389404, | |
| "step": 2555, | |
| "valid_targets_mean": 4201.8, | |
| "valid_targets_min": 1546 | |
| }, | |
| { | |
| "epoch": 4.579606440071556, | |
| "grad_norm": 0.5096054278407313, | |
| "learning_rate": 1.290405196472064e-05, | |
| "loss": 0.265, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26317811012268066, | |
| "step": 2560, | |
| "valid_targets_mean": 4815.1, | |
| "valid_targets_min": 1171 | |
| }, | |
| { | |
| "epoch": 4.588550983899821, | |
| "grad_norm": 0.5275387474839849, | |
| "learning_rate": 1.282070306742222e-05, | |
| "loss": 0.265, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2525607645511627, | |
| "step": 2565, | |
| "valid_targets_mean": 4492.8, | |
| "valid_targets_min": 1887 | |
| }, | |
| { | |
| "epoch": 4.597495527728086, | |
| "grad_norm": 0.5924005743194193, | |
| "learning_rate": 1.2737497055867482e-05, | |
| "loss": 0.2701, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2749042510986328, | |
| "step": 2570, | |
| "valid_targets_mean": 3256.1, | |
| "valid_targets_min": 825 | |
| }, | |
| { | |
| "epoch": 4.60644007155635, | |
| "grad_norm": 0.520307087440015, | |
| "learning_rate": 1.2654435586061604e-05, | |
| "loss": 0.2489, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23750732839107513, | |
| "step": 2575, | |
| "valid_targets_mean": 3960.6, | |
| "valid_targets_min": 1329 | |
| }, | |
| { | |
| "epoch": 4.615384615384615, | |
| "grad_norm": 0.5616936308653787, | |
| "learning_rate": 1.2571520311133012e-05, | |
| "loss": 0.2627, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26292529702186584, | |
| "step": 2580, | |
| "valid_targets_mean": 4504.7, | |
| "valid_targets_min": 1275 | |
| }, | |
| { | |
| "epoch": 4.624329159212881, | |
| "grad_norm": 0.5518910049186732, | |
| "learning_rate": 1.2488752881300514e-05, | |
| "loss": 0.2539, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2535983920097351, | |
| "step": 2585, | |
| "valid_targets_mean": 3939.8, | |
| "valid_targets_min": 1386 | |
| }, | |
| { | |
| "epoch": 4.633273703041145, | |
| "grad_norm": 0.5256431179864964, | |
| "learning_rate": 1.2406134943840398e-05, | |
| "loss": 0.2496, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2522655129432678, | |
| "step": 2590, | |
| "valid_targets_mean": 4140.9, | |
| "valid_targets_min": 474 | |
| }, | |
| { | |
| "epoch": 4.64221824686941, | |
| "grad_norm": 0.5048588742383056, | |
| "learning_rate": 1.232366814305372e-05, | |
| "loss": 0.2498, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24697677791118622, | |
| "step": 2595, | |
| "valid_targets_mean": 4505.9, | |
| "valid_targets_min": 1820 | |
| }, | |
| { | |
| "epoch": 4.651162790697675, | |
| "grad_norm": 0.6135798826851775, | |
| "learning_rate": 1.2241354120233527e-05, | |
| "loss": 0.2609, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27449357509613037, | |
| "step": 2600, | |
| "valid_targets_mean": 3534.1, | |
| "valid_targets_min": 1231 | |
| }, | |
| { | |
| "epoch": 4.660107334525939, | |
| "grad_norm": 0.5407420740372647, | |
| "learning_rate": 1.2159194513632213e-05, | |
| "loss": 0.2619, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2621440589427948, | |
| "step": 2605, | |
| "valid_targets_mean": 3698.6, | |
| "valid_targets_min": 625 | |
| }, | |
| { | |
| "epoch": 4.669051878354204, | |
| "grad_norm": 0.5784897601072413, | |
| "learning_rate": 1.20771909584289e-05, | |
| "loss": 0.2683, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2809554934501648, | |
| "step": 2610, | |
| "valid_targets_mean": 4114.5, | |
| "valid_targets_min": 1649 | |
| }, | |
| { | |
| "epoch": 4.677996422182469, | |
| "grad_norm": 0.5583707405753705, | |
| "learning_rate": 1.1995345086696919e-05, | |
| "loss": 0.2593, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.265133798122406, | |
| "step": 2615, | |
| "valid_targets_mean": 4105.6, | |
| "valid_targets_min": 1769 | |
| }, | |
| { | |
| "epoch": 4.6869409660107335, | |
| "grad_norm": 0.5899963254378155, | |
| "learning_rate": 1.1913658527371284e-05, | |
| "loss": 0.2642, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2800382077693939, | |
| "step": 2620, | |
| "valid_targets_mean": 3958.2, | |
| "valid_targets_min": 1014 | |
| }, | |
| { | |
| "epoch": 4.695885509838998, | |
| "grad_norm": 0.5666305355030801, | |
| "learning_rate": 1.1832132906216314e-05, | |
| "loss": 0.2681, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27610641717910767, | |
| "step": 2625, | |
| "valid_targets_mean": 3627.1, | |
| "valid_targets_min": 734 | |
| }, | |
| { | |
| "epoch": 4.704830053667263, | |
| "grad_norm": 0.48544029558526913, | |
| "learning_rate": 1.1750769845793268e-05, | |
| "loss": 0.2629, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25514447689056396, | |
| "step": 2630, | |
| "valid_targets_mean": 4717.1, | |
| "valid_targets_min": 767 | |
| }, | |
| { | |
| "epoch": 4.713774597495528, | |
| "grad_norm": 0.5191718129231461, | |
| "learning_rate": 1.1669570965428028e-05, | |
| "loss": 0.2708, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28754234313964844, | |
| "step": 2635, | |
| "valid_targets_mean": 4340.2, | |
| "valid_targets_min": 1685 | |
| }, | |
| { | |
| "epoch": 4.722719141323792, | |
| "grad_norm": 0.5824564729383654, | |
| "learning_rate": 1.1588537881178902e-05, | |
| "loss": 0.2545, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.259286105632782, | |
| "step": 2640, | |
| "valid_targets_mean": 3709.2, | |
| "valid_targets_min": 1083 | |
| }, | |
| { | |
| "epoch": 4.731663685152057, | |
| "grad_norm": 0.6199061316983084, | |
| "learning_rate": 1.150767220580444e-05, | |
| "loss": 0.2628, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.264919638633728, | |
| "step": 2645, | |
| "valid_targets_mean": 3604.6, | |
| "valid_targets_min": 1544 | |
| }, | |
| { | |
| "epoch": 4.740608228980322, | |
| "grad_norm": 0.4746750054946057, | |
| "learning_rate": 1.1426975548731329e-05, | |
| "loss": 0.2587, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21649254858493805, | |
| "step": 2650, | |
| "valid_targets_mean": 4358.8, | |
| "valid_targets_min": 1497 | |
| }, | |
| { | |
| "epoch": 4.7495527728085865, | |
| "grad_norm": 0.6094545797228554, | |
| "learning_rate": 1.1346449516022395e-05, | |
| "loss": 0.2525, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26640570163726807, | |
| "step": 2655, | |
| "valid_targets_mean": 3034.4, | |
| "valid_targets_min": 529 | |
| }, | |
| { | |
| "epoch": 4.758497316636851, | |
| "grad_norm": 0.5341381722241784, | |
| "learning_rate": 1.1266095710344611e-05, | |
| "loss": 0.261, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2599736154079437, | |
| "step": 2660, | |
| "valid_targets_mean": 4056.6, | |
| "valid_targets_min": 748 | |
| }, | |
| { | |
| "epoch": 4.767441860465116, | |
| "grad_norm": 0.5450354818964038, | |
| "learning_rate": 1.1185915730937209e-05, | |
| "loss": 0.2554, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2556326985359192, | |
| "step": 2665, | |
| "valid_targets_mean": 3869.1, | |
| "valid_targets_min": 1724 | |
| }, | |
| { | |
| "epoch": 4.7763864042933815, | |
| "grad_norm": 0.5823841949915403, | |
| "learning_rate": 1.1105911173579831e-05, | |
| "loss": 0.2757, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2735878527164459, | |
| "step": 2670, | |
| "valid_targets_mean": 3594.8, | |
| "valid_targets_min": 1608 | |
| }, | |
| { | |
| "epoch": 4.785330948121646, | |
| "grad_norm": 0.5282126999205208, | |
| "learning_rate": 1.1026083630560814e-05, | |
| "loss": 0.2526, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26514312624931335, | |
| "step": 2675, | |
| "valid_targets_mean": 4371.6, | |
| "valid_targets_min": 1156 | |
| }, | |
| { | |
| "epoch": 4.794275491949911, | |
| "grad_norm": 0.5463930991787092, | |
| "learning_rate": 1.0946434690645446e-05, | |
| "loss": 0.2611, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2594326138496399, | |
| "step": 2680, | |
| "valid_targets_mean": 4857.9, | |
| "valid_targets_min": 2646 | |
| }, | |
| { | |
| "epoch": 4.803220035778176, | |
| "grad_norm": 0.4924897165473573, | |
| "learning_rate": 1.08669659390444e-05, | |
| "loss": 0.2654, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2588515281677246, | |
| "step": 2685, | |
| "valid_targets_mean": 4642.6, | |
| "valid_targets_min": 2325 | |
| }, | |
| { | |
| "epoch": 4.81216457960644, | |
| "grad_norm": 0.4976503303623363, | |
| "learning_rate": 1.0787678957382124e-05, | |
| "loss": 0.2516, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.268010675907135, | |
| "step": 2690, | |
| "valid_targets_mean": 4382.6, | |
| "valid_targets_min": 1666 | |
| }, | |
| { | |
| "epoch": 4.821109123434705, | |
| "grad_norm": 0.5302669946862639, | |
| "learning_rate": 1.0708575323665422e-05, | |
| "loss": 0.2561, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25757157802581787, | |
| "step": 2695, | |
| "valid_targets_mean": 3814.4, | |
| "valid_targets_min": 1832 | |
| }, | |
| { | |
| "epoch": 4.83005366726297, | |
| "grad_norm": 0.546113045209434, | |
| "learning_rate": 1.062965661225201e-05, | |
| "loss": 0.2636, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2852606773376465, | |
| "step": 2700, | |
| "valid_targets_mean": 4466.5, | |
| "valid_targets_min": 1706 | |
| }, | |
| { | |
| "epoch": 4.838998211091234, | |
| "grad_norm": 0.508464282119227, | |
| "learning_rate": 1.0550924393819204e-05, | |
| "loss": 0.2638, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2477300763130188, | |
| "step": 2705, | |
| "valid_targets_mean": 4274.6, | |
| "valid_targets_min": 1461 | |
| }, | |
| { | |
| "epoch": 4.847942754919499, | |
| "grad_norm": 0.5908025845080628, | |
| "learning_rate": 1.0472380235332631e-05, | |
| "loss": 0.2632, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26860788464546204, | |
| "step": 2710, | |
| "valid_targets_mean": 3469.6, | |
| "valid_targets_min": 1287 | |
| }, | |
| { | |
| "epoch": 4.856887298747764, | |
| "grad_norm": 0.49449536043979037, | |
| "learning_rate": 1.0394025700015064e-05, | |
| "loss": 0.2601, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2539494037628174, | |
| "step": 2715, | |
| "valid_targets_mean": 5158.5, | |
| "valid_targets_min": 1956 | |
| }, | |
| { | |
| "epoch": 4.8658318425760285, | |
| "grad_norm": 0.5211386908969953, | |
| "learning_rate": 1.0315862347315314e-05, | |
| "loss": 0.263, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26820501685142517, | |
| "step": 2720, | |
| "valid_targets_mean": 4476.4, | |
| "valid_targets_min": 1901 | |
| }, | |
| { | |
| "epoch": 4.874776386404293, | |
| "grad_norm": 0.5544257720104306, | |
| "learning_rate": 1.0237891732877202e-05, | |
| "loss": 0.2671, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2941119372844696, | |
| "step": 2725, | |
| "valid_targets_mean": 4002.2, | |
| "valid_targets_min": 1560 | |
| }, | |
| { | |
| "epoch": 4.883720930232558, | |
| "grad_norm": 0.5465940219798077, | |
| "learning_rate": 1.0160115408508533e-05, | |
| "loss": 0.2693, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26257503032684326, | |
| "step": 2730, | |
| "valid_targets_mean": 3682.9, | |
| "valid_targets_min": 950 | |
| }, | |
| { | |
| "epoch": 4.892665474060823, | |
| "grad_norm": 0.5476342559629205, | |
| "learning_rate": 1.008253492215031e-05, | |
| "loss": 0.2565, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2684975862503052, | |
| "step": 2735, | |
| "valid_targets_mean": 4095.6, | |
| "valid_targets_min": 1925 | |
| }, | |
| { | |
| "epoch": 4.901610017889087, | |
| "grad_norm": 0.5180846052234739, | |
| "learning_rate": 1.0005151817845853e-05, | |
| "loss": 0.2729, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2808322012424469, | |
| "step": 2740, | |
| "valid_targets_mean": 4175.7, | |
| "valid_targets_min": 2464 | |
| }, | |
| { | |
| "epoch": 4.910554561717352, | |
| "grad_norm": 0.5936981738189633, | |
| "learning_rate": 9.927967635710087e-06, | |
| "loss": 0.2516, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24229755997657776, | |
| "step": 2745, | |
| "valid_targets_mean": 4147.9, | |
| "valid_targets_min": 914 | |
| }, | |
| { | |
| "epoch": 4.919499105545617, | |
| "grad_norm": 0.5464370878938494, | |
| "learning_rate": 9.850983911898913e-06, | |
| "loss": 0.2657, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24679221212863922, | |
| "step": 2750, | |
| "valid_targets_mean": 3841.9, | |
| "valid_targets_min": 877 | |
| }, | |
| { | |
| "epoch": 4.928443649373882, | |
| "grad_norm": 0.5334187085612581, | |
| "learning_rate": 9.774202178578592e-06, | |
| "loss": 0.2531, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2615172863006592, | |
| "step": 2755, | |
| "valid_targets_mean": 3857.0, | |
| "valid_targets_min": 622 | |
| }, | |
| { | |
| "epoch": 4.937388193202147, | |
| "grad_norm": 0.5666627948297285, | |
| "learning_rate": 9.69762396389529e-06, | |
| "loss": 0.2637, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2589084804058075, | |
| "step": 2760, | |
| "valid_targets_mean": 3624.9, | |
| "valid_targets_min": 1519 | |
| }, | |
| { | |
| "epoch": 4.946332737030412, | |
| "grad_norm": 0.5273284504363817, | |
| "learning_rate": 9.621250791944659e-06, | |
| "loss": 0.2627, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2510490417480469, | |
| "step": 2765, | |
| "valid_targets_mean": 5004.1, | |
| "valid_targets_min": 1979 | |
| }, | |
| { | |
| "epoch": 4.9552772808586765, | |
| "grad_norm": 0.6220423946842822, | |
| "learning_rate": 9.545084182741476e-06, | |
| "loss": 0.2723, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2961236536502838, | |
| "step": 2770, | |
| "valid_targets_mean": 3909.4, | |
| "valid_targets_min": 657 | |
| }, | |
| { | |
| "epoch": 4.964221824686941, | |
| "grad_norm": 0.6418856659491654, | |
| "learning_rate": 9.469125652189403e-06, | |
| "loss": 0.2672, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28539496660232544, | |
| "step": 2775, | |
| "valid_targets_mean": 3030.6, | |
| "valid_targets_min": 575 | |
| }, | |
| { | |
| "epoch": 4.973166368515206, | |
| "grad_norm": 0.5309277403670316, | |
| "learning_rate": 9.393376712050847e-06, | |
| "loss": 0.2637, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2643331289291382, | |
| "step": 2780, | |
| "valid_targets_mean": 4088.8, | |
| "valid_targets_min": 848 | |
| }, | |
| { | |
| "epoch": 4.982110912343471, | |
| "grad_norm": 0.6124698201076337, | |
| "learning_rate": 9.317838869916835e-06, | |
| "loss": 0.2632, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28254374861717224, | |
| "step": 2785, | |
| "valid_targets_mean": 3677.6, | |
| "valid_targets_min": 1521 | |
| }, | |
| { | |
| "epoch": 4.991055456171735, | |
| "grad_norm": 0.7441462785686875, | |
| "learning_rate": 9.242513629177033e-06, | |
| "loss": 0.261, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27683040499687195, | |
| "step": 2790, | |
| "valid_targets_mean": 3313.4, | |
| "valid_targets_min": 627 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 0.5230399259211697, | |
| "learning_rate": 9.167402488989784e-06, | |
| "loss": 0.2555, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25869959592819214, | |
| "step": 2795, | |
| "valid_targets_mean": 3854.7, | |
| "valid_targets_min": 743 | |
| }, | |
| { | |
| "epoch": 5.008944543828265, | |
| "grad_norm": 0.5549552542784999, | |
| "learning_rate": 9.092506944252344e-06, | |
| "loss": 0.2525, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2471010535955429, | |
| "step": 2800, | |
| "valid_targets_mean": 4048.8, | |
| "valid_targets_min": 1320 | |
| }, | |
| { | |
| "epoch": 5.017889087656529, | |
| "grad_norm": 0.5147728124256338, | |
| "learning_rate": 9.017828485571044e-06, | |
| "loss": 0.2399, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23798583447933197, | |
| "step": 2805, | |
| "valid_targets_mean": 4106.8, | |
| "valid_targets_min": 937 | |
| }, | |
| { | |
| "epoch": 5.026833631484794, | |
| "grad_norm": 0.5315317178738851, | |
| "learning_rate": 8.943368599231706e-06, | |
| "loss": 0.2644, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2732792794704437, | |
| "step": 2810, | |
| "valid_targets_mean": 4333.1, | |
| "valid_targets_min": 2013 | |
| }, | |
| { | |
| "epoch": 5.035778175313059, | |
| "grad_norm": 0.5729446923576372, | |
| "learning_rate": 8.869128767169986e-06, | |
| "loss": 0.2424, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2487604320049286, | |
| "step": 2815, | |
| "valid_targets_mean": 4101.1, | |
| "valid_targets_min": 1705 | |
| }, | |
| { | |
| "epoch": 5.0447227191413235, | |
| "grad_norm": 0.5034079404779742, | |
| "learning_rate": 8.79511046694194e-06, | |
| "loss": 0.2608, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26576727628707886, | |
| "step": 2820, | |
| "valid_targets_mean": 5089.2, | |
| "valid_targets_min": 2511 | |
| }, | |
| { | |
| "epoch": 5.053667262969588, | |
| "grad_norm": 0.6023654856445056, | |
| "learning_rate": 8.721315171694591e-06, | |
| "loss": 0.2452, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25126317143440247, | |
| "step": 2825, | |
| "valid_targets_mean": 3496.4, | |
| "valid_targets_min": 690 | |
| }, | |
| { | |
| "epoch": 5.062611806797853, | |
| "grad_norm": 0.6689786638114392, | |
| "learning_rate": 8.647744350136612e-06, | |
| "loss": 0.2462, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24577105045318604, | |
| "step": 2830, | |
| "valid_targets_mean": 2972.4, | |
| "valid_targets_min": 732 | |
| }, | |
| { | |
| "epoch": 5.071556350626118, | |
| "grad_norm": 0.5438242397653636, | |
| "learning_rate": 8.574399466509085e-06, | |
| "loss": 0.2551, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26564639806747437, | |
| "step": 2835, | |
| "valid_targets_mean": 3876.6, | |
| "valid_targets_min": 1618 | |
| }, | |
| { | |
| "epoch": 5.080500894454383, | |
| "grad_norm": 0.6404762750457129, | |
| "learning_rate": 8.501281980556369e-06, | |
| "loss": 0.2638, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2322574257850647, | |
| "step": 2840, | |
| "valid_targets_mean": 3335.4, | |
| "valid_targets_min": 1755 | |
| }, | |
| { | |
| "epoch": 5.089445438282648, | |
| "grad_norm": 0.5158771438098722, | |
| "learning_rate": 8.428393347497057e-06, | |
| "loss": 0.2549, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2659394145011902, | |
| "step": 2845, | |
| "valid_targets_mean": 4493.7, | |
| "valid_targets_min": 2239 | |
| }, | |
| { | |
| "epoch": 5.098389982110913, | |
| "grad_norm": 0.563631894055807, | |
| "learning_rate": 8.355735017995011e-06, | |
| "loss": 0.2518, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23874777555465698, | |
| "step": 2850, | |
| "valid_targets_mean": 3666.8, | |
| "valid_targets_min": 754 | |
| }, | |
| { | |
| "epoch": 5.107334525939177, | |
| "grad_norm": 0.5264091725894624, | |
| "learning_rate": 8.283308438130458e-06, | |
| "loss": 0.2557, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28656524419784546, | |
| "step": 2855, | |
| "valid_targets_mean": 4860.7, | |
| "valid_targets_min": 1880 | |
| }, | |
| { | |
| "epoch": 5.116279069767442, | |
| "grad_norm": 0.5531684984629897, | |
| "learning_rate": 8.211115049371266e-06, | |
| "loss": 0.2478, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25294798612594604, | |
| "step": 2860, | |
| "valid_targets_mean": 4066.2, | |
| "valid_targets_min": 634 | |
| }, | |
| { | |
| "epoch": 5.125223613595707, | |
| "grad_norm": 0.5887989735486293, | |
| "learning_rate": 8.13915628854419e-06, | |
| "loss": 0.246, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27043190598487854, | |
| "step": 2865, | |
| "valid_targets_mean": 3563.4, | |
| "valid_targets_min": 737 | |
| }, | |
| { | |
| "epoch": 5.1341681574239715, | |
| "grad_norm": 0.6198216758291454, | |
| "learning_rate": 8.067433587806336e-06, | |
| "loss": 0.2588, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2664569318294525, | |
| "step": 2870, | |
| "valid_targets_mean": 3196.0, | |
| "valid_targets_min": 646 | |
| }, | |
| { | |
| "epoch": 5.143112701252236, | |
| "grad_norm": 0.49039197556547504, | |
| "learning_rate": 7.995948374616628e-06, | |
| "loss": 0.2449, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24738174676895142, | |
| "step": 2875, | |
| "valid_targets_mean": 4813.1, | |
| "valid_targets_min": 1748 | |
| }, | |
| { | |
| "epoch": 5.152057245080501, | |
| "grad_norm": 0.5124249703499617, | |
| "learning_rate": 7.924702071707386e-06, | |
| "loss": 0.2482, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22936435043811798, | |
| "step": 2880, | |
| "valid_targets_mean": 4589.9, | |
| "valid_targets_min": 669 | |
| }, | |
| { | |
| "epoch": 5.161001788908766, | |
| "grad_norm": 0.5356467322668551, | |
| "learning_rate": 7.853696097056037e-06, | |
| "loss": 0.2582, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.264324426651001, | |
| "step": 2885, | |
| "valid_targets_mean": 4108.2, | |
| "valid_targets_min": 1605 | |
| }, | |
| { | |
| "epoch": 5.16994633273703, | |
| "grad_norm": 0.5087262397528598, | |
| "learning_rate": 7.782931863856888e-06, | |
| "loss": 0.248, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23341549932956696, | |
| "step": 2890, | |
| "valid_targets_mean": 4439.1, | |
| "valid_targets_min": 1726 | |
| }, | |
| { | |
| "epoch": 5.178890876565295, | |
| "grad_norm": 0.5549321487561256, | |
| "learning_rate": 7.71241078049298e-06, | |
| "loss": 0.2438, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25444790720939636, | |
| "step": 2895, | |
| "valid_targets_mean": 4032.2, | |
| "valid_targets_min": 1642 | |
| }, | |
| { | |
| "epoch": 5.18783542039356, | |
| "grad_norm": 0.5352806321855557, | |
| "learning_rate": 7.642134250508069e-06, | |
| "loss": 0.2452, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24765413999557495, | |
| "step": 2900, | |
| "valid_targets_mean": 4077.3, | |
| "valid_targets_min": 1757 | |
| }, | |
| { | |
| "epoch": 5.196779964221824, | |
| "grad_norm": 0.5518211501366974, | |
| "learning_rate": 7.57210367257871e-06, | |
| "loss": 0.2655, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24627630412578583, | |
| "step": 2905, | |
| "valid_targets_mean": 4041.1, | |
| "valid_targets_min": 1350 | |
| }, | |
| { | |
| "epoch": 5.205724508050089, | |
| "grad_norm": 0.5517560537112484, | |
| "learning_rate": 7.502320440486399e-06, | |
| "loss": 0.2516, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27135342359542847, | |
| "step": 2910, | |
| "valid_targets_mean": 4264.5, | |
| "valid_targets_min": 1184 | |
| }, | |
| { | |
| "epoch": 5.214669051878354, | |
| "grad_norm": 0.5509625444275201, | |
| "learning_rate": 7.4327859430898445e-06, | |
| "loss": 0.2573, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2660328149795532, | |
| "step": 2915, | |
| "valid_targets_mean": 4360.2, | |
| "valid_targets_min": 1551 | |
| }, | |
| { | |
| "epoch": 5.2236135957066185, | |
| "grad_norm": 0.5264725135537257, | |
| "learning_rate": 7.363501564297302e-06, | |
| "loss": 0.2503, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2478117048740387, | |
| "step": 2920, | |
| "valid_targets_mean": 4145.1, | |
| "valid_targets_min": 1811 | |
| }, | |
| { | |
| "epoch": 5.232558139534884, | |
| "grad_norm": 0.5015055063053451, | |
| "learning_rate": 7.294468683039058e-06, | |
| "loss": 0.2595, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24327683448791504, | |
| "step": 2925, | |
| "valid_targets_mean": 4711.1, | |
| "valid_targets_min": 714 | |
| }, | |
| { | |
| "epoch": 5.241502683363149, | |
| "grad_norm": 0.5747369854441176, | |
| "learning_rate": 7.225688673239981e-06, | |
| "loss": 0.2512, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24569492042064667, | |
| "step": 2930, | |
| "valid_targets_mean": 4338.2, | |
| "valid_targets_min": 687 | |
| }, | |
| { | |
| "epoch": 5.2504472271914135, | |
| "grad_norm": 0.558577994633376, | |
| "learning_rate": 7.1571629037921804e-06, | |
| "loss": 0.2526, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28280454874038696, | |
| "step": 2935, | |
| "valid_targets_mean": 4211.2, | |
| "valid_targets_min": 2352 | |
| }, | |
| { | |
| "epoch": 5.259391771019678, | |
| "grad_norm": 0.6144056807907267, | |
| "learning_rate": 7.088892738527731e-06, | |
| "loss": 0.2608, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2709640562534332, | |
| "step": 2940, | |
| "valid_targets_mean": 4119.1, | |
| "valid_targets_min": 1206 | |
| }, | |
| { | |
| "epoch": 5.268336314847943, | |
| "grad_norm": 0.5107272318888509, | |
| "learning_rate": 7.020879536191571e-06, | |
| "loss": 0.2497, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26789391040802, | |
| "step": 2945, | |
| "valid_targets_mean": 4640.5, | |
| "valid_targets_min": 1448 | |
| }, | |
| { | |
| "epoch": 5.277280858676208, | |
| "grad_norm": 0.5541009190577022, | |
| "learning_rate": 6.953124650414447e-06, | |
| "loss": 0.2512, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2534041404724121, | |
| "step": 2950, | |
| "valid_targets_mean": 4751.1, | |
| "valid_targets_min": 1213 | |
| }, | |
| { | |
| "epoch": 5.286225402504472, | |
| "grad_norm": 0.6514472569289943, | |
| "learning_rate": 6.885629429685943e-06, | |
| "loss": 0.2599, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2853846251964569, | |
| "step": 2955, | |
| "valid_targets_mean": 4381.8, | |
| "valid_targets_min": 1086 | |
| }, | |
| { | |
| "epoch": 5.295169946332737, | |
| "grad_norm": 0.5988322541910981, | |
| "learning_rate": 6.818395217327696e-06, | |
| "loss": 0.2577, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25575023889541626, | |
| "step": 2960, | |
| "valid_targets_mean": 3193.8, | |
| "valid_targets_min": 1521 | |
| }, | |
| { | |
| "epoch": 5.304114490161002, | |
| "grad_norm": 0.5670506850205758, | |
| "learning_rate": 6.751423351466604e-06, | |
| "loss": 0.2441, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23857812583446503, | |
| "step": 2965, | |
| "valid_targets_mean": 4116.7, | |
| "valid_targets_min": 2343 | |
| }, | |
| { | |
| "epoch": 5.3130590339892665, | |
| "grad_norm": 0.5218993251194615, | |
| "learning_rate": 6.684715165008242e-06, | |
| "loss": 0.2511, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24250607192516327, | |
| "step": 2970, | |
| "valid_targets_mean": 4213.7, | |
| "valid_targets_min": 712 | |
| }, | |
| { | |
| "epoch": 5.322003577817531, | |
| "grad_norm": 0.5569832821022914, | |
| "learning_rate": 6.618271985610316e-06, | |
| "loss": 0.2516, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2601429224014282, | |
| "step": 2975, | |
| "valid_targets_mean": 4093.6, | |
| "valid_targets_min": 1465 | |
| }, | |
| { | |
| "epoch": 5.330948121645796, | |
| "grad_norm": 0.5257447628225279, | |
| "learning_rate": 6.552095135656211e-06, | |
| "loss": 0.252, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22466397285461426, | |
| "step": 2980, | |
| "valid_targets_mean": 4426.8, | |
| "valid_targets_min": 1498 | |
| }, | |
| { | |
| "epoch": 5.339892665474061, | |
| "grad_norm": 0.5288226075745489, | |
| "learning_rate": 6.486185932228732e-06, | |
| "loss": 0.2495, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23786196112632751, | |
| "step": 2985, | |
| "valid_targets_mean": 4234.4, | |
| "valid_targets_min": 947 | |
| }, | |
| { | |
| "epoch": 5.348837209302325, | |
| "grad_norm": 0.5316983597863091, | |
| "learning_rate": 6.420545687083823e-06, | |
| "loss": 0.2453, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2561112642288208, | |
| "step": 2990, | |
| "valid_targets_mean": 4393.5, | |
| "valid_targets_min": 1639 | |
| }, | |
| { | |
| "epoch": 5.35778175313059, | |
| "grad_norm": 0.6336255850890024, | |
| "learning_rate": 6.355175706624521e-06, | |
| "loss": 0.2516, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24200129508972168, | |
| "step": 2995, | |
| "valid_targets_mean": 2892.8, | |
| "valid_targets_min": 1724 | |
| }, | |
| { | |
| "epoch": 5.366726296958855, | |
| "grad_norm": 0.5790999887623745, | |
| "learning_rate": 6.290077291874923e-06, | |
| "loss": 0.2472, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25445109605789185, | |
| "step": 3000, | |
| "valid_targets_mean": 3687.9, | |
| "valid_targets_min": 1561 | |
| }, | |
| { | |
| "epoch": 5.375670840787119, | |
| "grad_norm": 0.6224285293408105, | |
| "learning_rate": 6.225251738454281e-06, | |
| "loss": 0.2558, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2375735342502594, | |
| "step": 3005, | |
| "valid_targets_mean": 3882.3, | |
| "valid_targets_min": 1924 | |
| }, | |
| { | |
| "epoch": 5.384615384615385, | |
| "grad_norm": 0.5037477615762228, | |
| "learning_rate": 6.160700336551246e-06, | |
| "loss": 0.2484, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22593946754932404, | |
| "step": 3010, | |
| "valid_targets_mean": 4476.9, | |
| "valid_targets_min": 1419 | |
| }, | |
| { | |
| "epoch": 5.39355992844365, | |
| "grad_norm": 0.5857014550270156, | |
| "learning_rate": 6.096424370898186e-06, | |
| "loss": 0.2563, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26964280009269714, | |
| "step": 3015, | |
| "valid_targets_mean": 3775.2, | |
| "valid_targets_min": 942 | |
| }, | |
| { | |
| "epoch": 5.402504472271914, | |
| "grad_norm": 0.5658229988053839, | |
| "learning_rate": 6.032425120745586e-06, | |
| "loss": 0.255, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2709835171699524, | |
| "step": 3020, | |
| "valid_targets_mean": 3667.1, | |
| "valid_targets_min": 1957 | |
| }, | |
| { | |
| "epoch": 5.411449016100179, | |
| "grad_norm": 0.5988261907445214, | |
| "learning_rate": 5.968703859836615e-06, | |
| "loss": 0.2562, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2751755118370056, | |
| "step": 3025, | |
| "valid_targets_mean": 3496.3, | |
| "valid_targets_min": 671 | |
| }, | |
| { | |
| "epoch": 5.420393559928444, | |
| "grad_norm": 0.5141241348502327, | |
| "learning_rate": 5.905261856381779e-06, | |
| "loss": 0.257, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2791970372200012, | |
| "step": 3030, | |
| "valid_targets_mean": 4357.8, | |
| "valid_targets_min": 1171 | |
| }, | |
| { | |
| "epoch": 5.4293381037567086, | |
| "grad_norm": 0.5498788228367728, | |
| "learning_rate": 5.8421003730336635e-06, | |
| "loss": 0.2532, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2511694133281708, | |
| "step": 3035, | |
| "valid_targets_mean": 3824.6, | |
| "valid_targets_min": 1185 | |
| }, | |
| { | |
| "epoch": 5.438282647584973, | |
| "grad_norm": 0.6029343448549779, | |
| "learning_rate": 5.77922066686182e-06, | |
| "loss": 0.2482, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.243141770362854, | |
| "step": 3040, | |
| "valid_targets_mean": 3373.0, | |
| "valid_targets_min": 823 | |
| }, | |
| { | |
| "epoch": 5.447227191413238, | |
| "grad_norm": 0.5131367304880754, | |
| "learning_rate": 5.716623989327723e-06, | |
| "loss": 0.2504, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2467704713344574, | |
| "step": 3045, | |
| "valid_targets_mean": 4253.5, | |
| "valid_targets_min": 1483 | |
| }, | |
| { | |
| "epoch": 5.456171735241503, | |
| "grad_norm": 0.5805462950603849, | |
| "learning_rate": 5.654311586259882e-06, | |
| "loss": 0.25, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23988988995552063, | |
| "step": 3050, | |
| "valid_targets_mean": 4065.0, | |
| "valid_targets_min": 2083 | |
| }, | |
| { | |
| "epoch": 5.465116279069767, | |
| "grad_norm": 0.5715523149741332, | |
| "learning_rate": 5.592284697829049e-06, | |
| "loss": 0.2435, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2494080364704132, | |
| "step": 3055, | |
| "valid_targets_mean": 3334.1, | |
| "valid_targets_min": 734 | |
| }, | |
| { | |
| "epoch": 5.474060822898032, | |
| "grad_norm": 0.5721433275869938, | |
| "learning_rate": 5.5305445585235315e-06, | |
| "loss": 0.2598, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28742748498916626, | |
| "step": 3060, | |
| "valid_targets_mean": 3750.8, | |
| "valid_targets_min": 1385 | |
| }, | |
| { | |
| "epoch": 5.483005366726297, | |
| "grad_norm": 0.5114459072142161, | |
| "learning_rate": 5.469092397124609e-06, | |
| "loss": 0.2354, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2271006852388382, | |
| "step": 3065, | |
| "valid_targets_mean": 4397.1, | |
| "valid_targets_min": 1489 | |
| }, | |
| { | |
| "epoch": 5.4919499105545615, | |
| "grad_norm": 0.5521229518692201, | |
| "learning_rate": 5.407929436682098e-06, | |
| "loss": 0.2433, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25235363841056824, | |
| "step": 3070, | |
| "valid_targets_mean": 4136.6, | |
| "valid_targets_min": 1627 | |
| }, | |
| { | |
| "epoch": 5.500894454382826, | |
| "grad_norm": 0.5402233613552551, | |
| "learning_rate": 5.347056894490006e-06, | |
| "loss": 0.2514, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2618515193462372, | |
| "step": 3075, | |
| "valid_targets_mean": 4445.6, | |
| "valid_targets_min": 1844 | |
| }, | |
| { | |
| "epoch": 5.509838998211091, | |
| "grad_norm": 0.5533934913144476, | |
| "learning_rate": 5.286475982062283e-06, | |
| "loss": 0.2572, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26157045364379883, | |
| "step": 3080, | |
| "valid_targets_mean": 4031.0, | |
| "valid_targets_min": 1056 | |
| }, | |
| { | |
| "epoch": 5.518783542039356, | |
| "grad_norm": 0.5110038510091776, | |
| "learning_rate": 5.226187905108746e-06, | |
| "loss": 0.2628, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2799553871154785, | |
| "step": 3085, | |
| "valid_targets_mean": 4822.0, | |
| "valid_targets_min": 1524 | |
| }, | |
| { | |
| "epoch": 5.52772808586762, | |
| "grad_norm": 0.6041299965242917, | |
| "learning_rate": 5.166193863511046e-06, | |
| "loss": 0.2478, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25382912158966064, | |
| "step": 3090, | |
| "valid_targets_mean": 4463.2, | |
| "valid_targets_min": 1625 | |
| }, | |
| { | |
| "epoch": 5.536672629695886, | |
| "grad_norm": 0.523055625256692, | |
| "learning_rate": 5.106495051298815e-06, | |
| "loss": 0.2601, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26213690638542175, | |
| "step": 3095, | |
| "valid_targets_mean": 4872.1, | |
| "valid_targets_min": 1339 | |
| }, | |
| { | |
| "epoch": 5.545617173524151, | |
| "grad_norm": 0.689189142486139, | |
| "learning_rate": 5.047092656625887e-06, | |
| "loss": 0.2464, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22535747289657593, | |
| "step": 3100, | |
| "valid_targets_mean": 4960.2, | |
| "valid_targets_min": 1725 | |
| }, | |
| { | |
| "epoch": 5.554561717352415, | |
| "grad_norm": 0.5703961451507539, | |
| "learning_rate": 4.98798786174667e-06, | |
| "loss": 0.2567, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23777396976947784, | |
| "step": 3105, | |
| "valid_targets_mean": 3530.8, | |
| "valid_targets_min": 1099 | |
| }, | |
| { | |
| "epoch": 5.56350626118068, | |
| "grad_norm": 0.5659413248498028, | |
| "learning_rate": 4.9291818429925585e-06, | |
| "loss": 0.2503, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2758846580982208, | |
| "step": 3110, | |
| "valid_targets_mean": 3740.4, | |
| "valid_targets_min": 1606 | |
| }, | |
| { | |
| "epoch": 5.572450805008945, | |
| "grad_norm": 0.6859304017050197, | |
| "learning_rate": 4.8706757707486075e-06, | |
| "loss": 0.2556, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24378885328769684, | |
| "step": 3115, | |
| "valid_targets_mean": 4183.4, | |
| "valid_targets_min": 1253 | |
| }, | |
| { | |
| "epoch": 5.5813953488372094, | |
| "grad_norm": 0.8010245971038674, | |
| "learning_rate": 4.812470809430179e-06, | |
| "loss": 0.2506, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2438773214817047, | |
| "step": 3120, | |
| "valid_targets_mean": 4535.4, | |
| "valid_targets_min": 1654 | |
| }, | |
| { | |
| "epoch": 5.590339892665474, | |
| "grad_norm": 0.48380553908118173, | |
| "learning_rate": 4.754568117459788e-06, | |
| "loss": 0.2514, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23566769063472748, | |
| "step": 3125, | |
| "valid_targets_mean": 4967.9, | |
| "valid_targets_min": 2282 | |
| }, | |
| { | |
| "epoch": 5.599284436493739, | |
| "grad_norm": 0.555628362724338, | |
| "learning_rate": 4.69696884724403e-06, | |
| "loss": 0.2587, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26789629459381104, | |
| "step": 3130, | |
| "valid_targets_mean": 3930.6, | |
| "valid_targets_min": 1612 | |
| }, | |
| { | |
| "epoch": 5.608228980322004, | |
| "grad_norm": 0.5726463627079651, | |
| "learning_rate": 4.639674145150679e-06, | |
| "loss": 0.2703, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27078795433044434, | |
| "step": 3135, | |
| "valid_targets_mean": 4010.0, | |
| "valid_targets_min": 2215 | |
| }, | |
| { | |
| "epoch": 5.617173524150268, | |
| "grad_norm": 0.5160136519489497, | |
| "learning_rate": 4.582685151485826e-06, | |
| "loss": 0.254, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2637634873390198, | |
| "step": 3140, | |
| "valid_targets_mean": 4663.4, | |
| "valid_targets_min": 1675 | |
| }, | |
| { | |
| "epoch": 5.626118067978533, | |
| "grad_norm": 0.5831172100665993, | |
| "learning_rate": 4.526003000471238e-06, | |
| "loss": 0.252, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25619733333587646, | |
| "step": 3145, | |
| "valid_targets_mean": 3704.6, | |
| "valid_targets_min": 1062 | |
| }, | |
| { | |
| "epoch": 5.635062611806798, | |
| "grad_norm": 0.5228549236039476, | |
| "learning_rate": 4.469628820221723e-06, | |
| "loss": 0.2481, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25507116317749023, | |
| "step": 3150, | |
| "valid_targets_mean": 4272.4, | |
| "valid_targets_min": 1740 | |
| }, | |
| { | |
| "epoch": 5.644007155635062, | |
| "grad_norm": 0.5442853215358627, | |
| "learning_rate": 4.4135637327227365e-06, | |
| "loss": 0.2489, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24775390326976776, | |
| "step": 3155, | |
| "valid_targets_mean": 4432.6, | |
| "valid_targets_min": 2006 | |
| }, | |
| { | |
| "epoch": 5.652951699463327, | |
| "grad_norm": 0.6205216192517241, | |
| "learning_rate": 4.357808853808014e-06, | |
| "loss": 0.2484, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26101428270339966, | |
| "step": 3160, | |
| "valid_targets_mean": 3737.1, | |
| "valid_targets_min": 644 | |
| }, | |
| { | |
| "epoch": 5.661896243291592, | |
| "grad_norm": 0.5132795398550757, | |
| "learning_rate": 4.302365293137383e-06, | |
| "loss": 0.2558, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2703346014022827, | |
| "step": 3165, | |
| "valid_targets_mean": 4475.0, | |
| "valid_targets_min": 1557 | |
| }, | |
| { | |
| "epoch": 5.6708407871198565, | |
| "grad_norm": 0.5446296655760452, | |
| "learning_rate": 4.247234154174649e-06, | |
| "loss": 0.2565, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.261799693107605, | |
| "step": 3170, | |
| "valid_targets_mean": 4172.8, | |
| "valid_targets_min": 823 | |
| }, | |
| { | |
| "epoch": 5.679785330948121, | |
| "grad_norm": 0.5219059793104994, | |
| "learning_rate": 4.192416534165666e-06, | |
| "loss": 0.2551, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2537573277950287, | |
| "step": 3175, | |
| "valid_targets_mean": 4267.9, | |
| "valid_targets_min": 579 | |
| }, | |
| { | |
| "epoch": 5.688729874776387, | |
| "grad_norm": 0.5461673557485848, | |
| "learning_rate": 4.137913524116488e-06, | |
| "loss": 0.2347, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2290881723165512, | |
| "step": 3180, | |
| "valid_targets_mean": 4211.2, | |
| "valid_targets_min": 818 | |
| }, | |
| { | |
| "epoch": 5.6976744186046515, | |
| "grad_norm": 0.517997717225711, | |
| "learning_rate": 4.083726208771653e-06, | |
| "loss": 0.2514, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25634273886680603, | |
| "step": 3185, | |
| "valid_targets_mean": 4610.6, | |
| "valid_targets_min": 1674 | |
| }, | |
| { | |
| "epoch": 5.706618962432916, | |
| "grad_norm": 0.5213271875253821, | |
| "learning_rate": 4.0298556665925996e-06, | |
| "loss": 0.2611, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27138853073120117, | |
| "step": 3190, | |
| "valid_targets_mean": 4529.8, | |
| "valid_targets_min": 1865 | |
| }, | |
| { | |
| "epoch": 5.715563506261181, | |
| "grad_norm": 0.546634737650849, | |
| "learning_rate": 3.976302969736183e-06, | |
| "loss": 0.2587, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2496567666530609, | |
| "step": 3195, | |
| "valid_targets_mean": 3815.2, | |
| "valid_targets_min": 932 | |
| }, | |
| { | |
| "epoch": 5.724508050089446, | |
| "grad_norm": 0.5258705612970683, | |
| "learning_rate": 3.923069184033357e-06, | |
| "loss": 0.2509, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23271816968917847, | |
| "step": 3200, | |
| "valid_targets_mean": 4088.2, | |
| "valid_targets_min": 2116 | |
| }, | |
| { | |
| "epoch": 5.73345259391771, | |
| "grad_norm": 0.5867279766106626, | |
| "learning_rate": 3.870155368967963e-06, | |
| "loss": 0.2536, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25471609830856323, | |
| "step": 3205, | |
| "valid_targets_mean": 3395.8, | |
| "valid_targets_min": 1808 | |
| }, | |
| { | |
| "epoch": 5.742397137745975, | |
| "grad_norm": 0.5656851680671992, | |
| "learning_rate": 3.8175625776556354e-06, | |
| "loss": 0.2578, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26453453302383423, | |
| "step": 3210, | |
| "valid_targets_mean": 4103.0, | |
| "valid_targets_min": 786 | |
| }, | |
| { | |
| "epoch": 5.75134168157424, | |
| "grad_norm": 0.5284899901213221, | |
| "learning_rate": 3.76529185682283e-06, | |
| "loss": 0.2562, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24859002232551575, | |
| "step": 3215, | |
| "valid_targets_mean": 4613.4, | |
| "valid_targets_min": 1303 | |
| }, | |
| { | |
| "epoch": 5.7602862254025045, | |
| "grad_norm": 0.5635253947920319, | |
| "learning_rate": 3.7133442467860215e-06, | |
| "loss": 0.2532, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24023880064487457, | |
| "step": 3220, | |
| "valid_targets_mean": 3926.2, | |
| "valid_targets_min": 822 | |
| }, | |
| { | |
| "epoch": 5.769230769230769, | |
| "grad_norm": 0.6683290106994736, | |
| "learning_rate": 3.6617207814309686e-06, | |
| "loss": 0.2535, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24666255712509155, | |
| "step": 3225, | |
| "valid_targets_mean": 3464.2, | |
| "valid_targets_min": 1681 | |
| }, | |
| { | |
| "epoch": 5.778175313059034, | |
| "grad_norm": 0.5711365259269587, | |
| "learning_rate": 3.610422488192169e-06, | |
| "loss": 0.2474, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22737815976142883, | |
| "step": 3230, | |
| "valid_targets_mean": 3654.8, | |
| "valid_targets_min": 1601 | |
| }, | |
| { | |
| "epoch": 5.787119856887299, | |
| "grad_norm": 0.5262566407141166, | |
| "learning_rate": 3.559450388032355e-06, | |
| "loss": 0.2405, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23528073728084564, | |
| "step": 3235, | |
| "valid_targets_mean": 4000.4, | |
| "valid_targets_min": 1795 | |
| }, | |
| { | |
| "epoch": 5.796064400715563, | |
| "grad_norm": 0.5857070524704964, | |
| "learning_rate": 3.508805495422247e-06, | |
| "loss": 0.2497, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25850462913513184, | |
| "step": 3240, | |
| "valid_targets_mean": 3548.4, | |
| "valid_targets_min": 2027 | |
| }, | |
| { | |
| "epoch": 5.805008944543828, | |
| "grad_norm": 0.5422739378299365, | |
| "learning_rate": 3.4584888183203134e-06, | |
| "loss": 0.2454, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23234635591506958, | |
| "step": 3245, | |
| "valid_targets_mean": 3974.6, | |
| "valid_targets_min": 1831 | |
| }, | |
| { | |
| "epoch": 5.813953488372093, | |
| "grad_norm": 0.5467138164186622, | |
| "learning_rate": 3.4085013581527295e-06, | |
| "loss": 0.2502, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25337696075439453, | |
| "step": 3250, | |
| "valid_targets_mean": 3679.7, | |
| "valid_targets_min": 1793 | |
| }, | |
| { | |
| "epoch": 5.822898032200357, | |
| "grad_norm": 0.5923423600872293, | |
| "learning_rate": 3.3588441097934223e-06, | |
| "loss": 0.2552, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27606481313705444, | |
| "step": 3255, | |
| "valid_targets_mean": 4041.6, | |
| "valid_targets_min": 1902 | |
| }, | |
| { | |
| "epoch": 5.831842576028622, | |
| "grad_norm": 0.544712722817465, | |
| "learning_rate": 3.30951806154431e-06, | |
| "loss": 0.2573, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24701544642448425, | |
| "step": 3260, | |
| "valid_targets_mean": 4252.6, | |
| "valid_targets_min": 1603 | |
| }, | |
| { | |
| "epoch": 5.840787119856888, | |
| "grad_norm": 0.5682646260679644, | |
| "learning_rate": 3.260524195115595e-06, | |
| "loss": 0.2527, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2743680477142334, | |
| "step": 3265, | |
| "valid_targets_mean": 4033.4, | |
| "valid_targets_min": 1567 | |
| }, | |
| { | |
| "epoch": 5.849731663685152, | |
| "grad_norm": 0.5053561115203649, | |
| "learning_rate": 3.2118634856062546e-06, | |
| "loss": 0.2515, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25195997953414917, | |
| "step": 3270, | |
| "valid_targets_mean": 4670.6, | |
| "valid_targets_min": 1711 | |
| }, | |
| { | |
| "epoch": 5.858676207513417, | |
| "grad_norm": 0.5720698542136006, | |
| "learning_rate": 3.163536901484601e-06, | |
| "loss": 0.2579, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25598081946372986, | |
| "step": 3275, | |
| "valid_targets_mean": 3693.4, | |
| "valid_targets_min": 850 | |
| }, | |
| { | |
| "epoch": 5.867620751341682, | |
| "grad_norm": 0.5504159278631617, | |
| "learning_rate": 3.115545404569045e-06, | |
| "loss": 0.2491, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2605506181716919, | |
| "step": 3280, | |
| "valid_targets_mean": 3985.3, | |
| "valid_targets_min": 1494 | |
| }, | |
| { | |
| "epoch": 5.8765652951699465, | |
| "grad_norm": 0.5073092549610757, | |
| "learning_rate": 3.0678899500089244e-06, | |
| "loss": 0.2593, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28743916749954224, | |
| "step": 3285, | |
| "valid_targets_mean": 5028.8, | |
| "valid_targets_min": 2241 | |
| }, | |
| { | |
| "epoch": 5.885509838998211, | |
| "grad_norm": 0.5340456565641353, | |
| "learning_rate": 3.0205714862655135e-06, | |
| "loss": 0.2539, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2738155722618103, | |
| "step": 3290, | |
| "valid_targets_mean": 4406.5, | |
| "valid_targets_min": 1014 | |
| }, | |
| { | |
| "epoch": 5.894454382826476, | |
| "grad_norm": 0.5891360814807866, | |
| "learning_rate": 2.973590955093124e-06, | |
| "loss": 0.2584, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26489949226379395, | |
| "step": 3295, | |
| "valid_targets_mean": 3631.4, | |
| "valid_targets_min": 1316 | |
| }, | |
| { | |
| "epoch": 5.903398926654741, | |
| "grad_norm": 0.606826380313794, | |
| "learning_rate": 2.9269492915203735e-06, | |
| "loss": 0.2528, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28875553607940674, | |
| "step": 3300, | |
| "valid_targets_mean": 3582.3, | |
| "valid_targets_min": 748 | |
| }, | |
| { | |
| "epoch": 5.912343470483005, | |
| "grad_norm": 0.5117519052350044, | |
| "learning_rate": 2.880647423831591e-06, | |
| "loss": 0.2548, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2642664909362793, | |
| "step": 3305, | |
| "valid_targets_mean": 5080.2, | |
| "valid_targets_min": 747 | |
| }, | |
| { | |
| "epoch": 5.92128801431127, | |
| "grad_norm": 0.5806379654175979, | |
| "learning_rate": 2.8346862735483215e-06, | |
| "loss": 0.2522, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2645312547683716, | |
| "step": 3310, | |
| "valid_targets_mean": 3762.3, | |
| "valid_targets_min": 1805 | |
| }, | |
| { | |
| "epoch": 5.930232558139535, | |
| "grad_norm": 0.5311509065424678, | |
| "learning_rate": 2.789066755410994e-06, | |
| "loss": 0.2488, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2585093379020691, | |
| "step": 3315, | |
| "valid_targets_mean": 4278.2, | |
| "valid_targets_min": 1326 | |
| }, | |
| { | |
| "epoch": 5.9391771019677995, | |
| "grad_norm": 0.5936285800316835, | |
| "learning_rate": 2.7437897773607102e-06, | |
| "loss": 0.2559, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26041656732559204, | |
| "step": 3320, | |
| "valid_targets_mean": 3442.7, | |
| "valid_targets_min": 1502 | |
| }, | |
| { | |
| "epoch": 5.948121645796064, | |
| "grad_norm": 0.5459969266201287, | |
| "learning_rate": 2.698856240521173e-06, | |
| "loss": 0.2566, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25794172286987305, | |
| "step": 3325, | |
| "valid_targets_mean": 4014.2, | |
| "valid_targets_min": 1551 | |
| }, | |
| { | |
| "epoch": 5.957066189624329, | |
| "grad_norm": 0.5630581255206843, | |
| "learning_rate": 2.6542670391807667e-06, | |
| "loss": 0.2533, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24545322358608246, | |
| "step": 3330, | |
| "valid_targets_mean": 3481.9, | |
| "valid_targets_min": 1557 | |
| }, | |
| { | |
| "epoch": 5.966010733452594, | |
| "grad_norm": 0.5799194413488256, | |
| "learning_rate": 2.610023060774749e-06, | |
| "loss": 0.2549, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2724427580833435, | |
| "step": 3335, | |
| "valid_targets_mean": 3878.4, | |
| "valid_targets_min": 702 | |
| }, | |
| { | |
| "epoch": 5.974955277280858, | |
| "grad_norm": 0.5740945458943522, | |
| "learning_rate": 2.566125185867576e-06, | |
| "loss": 0.2469, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23313261568546295, | |
| "step": 3340, | |
| "valid_targets_mean": 3677.9, | |
| "valid_targets_min": 923 | |
| }, | |
| { | |
| "epoch": 5.983899821109123, | |
| "grad_norm": 0.4939353323910504, | |
| "learning_rate": 2.5225742881354043e-06, | |
| "loss": 0.2455, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23637865483760834, | |
| "step": 3345, | |
| "valid_targets_mean": 5144.5, | |
| "valid_targets_min": 2196 | |
| }, | |
| { | |
| "epoch": 5.992844364937389, | |
| "grad_norm": 0.5709249769648037, | |
| "learning_rate": 2.479371234348682e-06, | |
| "loss": 0.2718, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2716667056083679, | |
| "step": 3350, | |
| "valid_targets_mean": 4413.6, | |
| "valid_targets_min": 1591 | |
| }, | |
| { | |
| "epoch": 6.001788908765653, | |
| "grad_norm": 0.49922306688966933, | |
| "learning_rate": 2.4365168843549e-06, | |
| "loss": 0.2583, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24928270280361176, | |
| "step": 3355, | |
| "valid_targets_mean": 4612.2, | |
| "valid_targets_min": 1207 | |
| }, | |
| { | |
| "epoch": 6.010733452593918, | |
| "grad_norm": 0.5555093196620423, | |
| "learning_rate": 2.3940120910614774e-06, | |
| "loss": 0.2536, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2676859498023987, | |
| "step": 3360, | |
| "valid_targets_mean": 3853.4, | |
| "valid_targets_min": 1449 | |
| }, | |
| { | |
| "epoch": 6.019677996422183, | |
| "grad_norm": 0.5566602842743498, | |
| "learning_rate": 2.351857700418807e-06, | |
| "loss": 0.241, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2452341467142105, | |
| "step": 3365, | |
| "valid_targets_mean": 4434.4, | |
| "valid_targets_min": 777 | |
| }, | |
| { | |
| "epoch": 6.028622540250447, | |
| "grad_norm": 0.5088695531600443, | |
| "learning_rate": 2.310054551403391e-06, | |
| "loss": 0.2387, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2538371682167053, | |
| "step": 3370, | |
| "valid_targets_mean": 4775.4, | |
| "valid_targets_min": 1902 | |
| }, | |
| { | |
| "epoch": 6.037567084078712, | |
| "grad_norm": 0.5558924991251052, | |
| "learning_rate": 2.2686034760011656e-06, | |
| "loss": 0.2435, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23813003301620483, | |
| "step": 3375, | |
| "valid_targets_mean": 3898.2, | |
| "valid_targets_min": 1674 | |
| }, | |
| { | |
| "epoch": 6.046511627906977, | |
| "grad_norm": 0.5591093005835225, | |
| "learning_rate": 2.2275052991909218e-06, | |
| "loss": 0.2417, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21672865748405457, | |
| "step": 3380, | |
| "valid_targets_mean": 4234.8, | |
| "valid_targets_min": 671 | |
| }, | |
| { | |
| "epoch": 6.0554561717352415, | |
| "grad_norm": 0.5260886376788583, | |
| "learning_rate": 2.186760838927897e-06, | |
| "loss": 0.2507, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22722241282463074, | |
| "step": 3385, | |
| "valid_targets_mean": 4410.5, | |
| "valid_targets_min": 575 | |
| }, | |
| { | |
| "epoch": 6.064400715563506, | |
| "grad_norm": 0.5613980727726008, | |
| "learning_rate": 2.146370906127506e-06, | |
| "loss": 0.2408, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2486397624015808, | |
| "step": 3390, | |
| "valid_targets_mean": 3953.5, | |
| "valid_targets_min": 1297 | |
| }, | |
| { | |
| "epoch": 6.073345259391771, | |
| "grad_norm": 0.5718364438676335, | |
| "learning_rate": 2.1063363046491813e-06, | |
| "loss": 0.2404, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24342885613441467, | |
| "step": 3395, | |
| "valid_targets_mean": 4326.2, | |
| "valid_targets_min": 1784 | |
| }, | |
| { | |
| "epoch": 6.082289803220036, | |
| "grad_norm": 0.5484733914155208, | |
| "learning_rate": 2.0666578312803942e-06, | |
| "loss": 0.2488, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23995618522167206, | |
| "step": 3400, | |
| "valid_targets_mean": 4243.8, | |
| "valid_targets_min": 1656 | |
| }, | |
| { | |
| "epoch": 6.0912343470483, | |
| "grad_norm": 0.6076003177346635, | |
| "learning_rate": 2.0273362757207726e-06, | |
| "loss": 0.2529, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2636367678642273, | |
| "step": 3405, | |
| "valid_targets_mean": 3371.1, | |
| "valid_targets_min": 700 | |
| }, | |
| { | |
| "epoch": 6.100178890876565, | |
| "grad_norm": 0.5364865052473711, | |
| "learning_rate": 1.988372420566416e-06, | |
| "loss": 0.2525, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2631525993347168, | |
| "step": 3410, | |
| "valid_targets_mean": 4552.2, | |
| "valid_targets_min": 527 | |
| }, | |
| { | |
| "epoch": 6.10912343470483, | |
| "grad_norm": 0.5382573273538706, | |
| "learning_rate": 1.9497670412942835e-06, | |
| "loss": 0.2434, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25336796045303345, | |
| "step": 3415, | |
| "valid_targets_mean": 4031.2, | |
| "valid_targets_min": 1500 | |
| }, | |
| { | |
| "epoch": 6.1180679785330945, | |
| "grad_norm": 0.5332393694448714, | |
| "learning_rate": 1.9115209062467954e-06, | |
| "loss": 0.248, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25488486886024475, | |
| "step": 3420, | |
| "valid_targets_mean": 4084.2, | |
| "valid_targets_min": 1073 | |
| }, | |
| { | |
| "epoch": 6.127012522361359, | |
| "grad_norm": 0.5822463874616248, | |
| "learning_rate": 1.8736347766165086e-06, | |
| "loss": 0.2415, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25610625743865967, | |
| "step": 3425, | |
| "valid_targets_mean": 3508.5, | |
| "valid_targets_min": 644 | |
| }, | |
| { | |
| "epoch": 6.135957066189624, | |
| "grad_norm": 0.516320931889104, | |
| "learning_rate": 1.8361094064309993e-06, | |
| "loss": 0.2523, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26783958077430725, | |
| "step": 3430, | |
| "valid_targets_mean": 4032.3, | |
| "valid_targets_min": 1287 | |
| }, | |
| { | |
| "epoch": 6.1449016100178895, | |
| "grad_norm": 0.5331137313762538, | |
| "learning_rate": 1.7989455425378266e-06, | |
| "loss": 0.2499, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2545079290866852, | |
| "step": 3435, | |
| "valid_targets_mean": 4148.0, | |
| "valid_targets_min": 1830 | |
| }, | |
| { | |
| "epoch": 6.153846153846154, | |
| "grad_norm": 0.511372820106071, | |
| "learning_rate": 1.7621439245896943e-06, | |
| "loss": 0.2476, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25283390283584595, | |
| "step": 3440, | |
| "valid_targets_mean": 4522.1, | |
| "valid_targets_min": 1614 | |
| }, | |
| { | |
| "epoch": 6.162790697674419, | |
| "grad_norm": 0.642569852208313, | |
| "learning_rate": 1.7257052850296996e-06, | |
| "loss": 0.2357, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24894089996814728, | |
| "step": 3445, | |
| "valid_targets_mean": 3752.3, | |
| "valid_targets_min": 697 | |
| }, | |
| { | |
| "epoch": 6.171735241502684, | |
| "grad_norm": 0.5666144603072086, | |
| "learning_rate": 1.6896303490767829e-06, | |
| "loss": 0.2418, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24932259321212769, | |
| "step": 3450, | |
| "valid_targets_mean": 3750.7, | |
| "valid_targets_min": 1707 | |
| }, | |
| { | |
| "epoch": 6.180679785330948, | |
| "grad_norm": 0.5143915126274258, | |
| "learning_rate": 1.6539198347112884e-06, | |
| "loss": 0.245, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2590116560459137, | |
| "step": 3455, | |
| "valid_targets_mean": 4465.8, | |
| "valid_targets_min": 1477 | |
| }, | |
| { | |
| "epoch": 6.189624329159213, | |
| "grad_norm": 0.5657080432140568, | |
| "learning_rate": 1.6185744526606706e-06, | |
| "loss": 0.2248, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22567050158977509, | |
| "step": 3460, | |
| "valid_targets_mean": 4094.5, | |
| "valid_targets_min": 1981 | |
| }, | |
| { | |
| "epoch": 6.198568872987478, | |
| "grad_norm": 0.6089266321722668, | |
| "learning_rate": 1.583594906385335e-06, | |
| "loss": 0.2516, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2387622594833374, | |
| "step": 3465, | |
| "valid_targets_mean": 4029.4, | |
| "valid_targets_min": 1928 | |
| }, | |
| { | |
| "epoch": 6.207513416815742, | |
| "grad_norm": 0.4852359479822391, | |
| "learning_rate": 1.548981892064665e-06, | |
| "loss": 0.2574, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24283546209335327, | |
| "step": 3470, | |
| "valid_targets_mean": 5261.4, | |
| "valid_targets_min": 1062 | |
| }, | |
| { | |
| "epoch": 6.216457960644007, | |
| "grad_norm": 0.4537685331486559, | |
| "learning_rate": 1.5147360985831516e-06, | |
| "loss": 0.2465, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21774128079414368, | |
| "step": 3475, | |
| "valid_targets_mean": 5369.9, | |
| "valid_targets_min": 3175 | |
| }, | |
| { | |
| "epoch": 6.225402504472272, | |
| "grad_norm": 0.5236024964731602, | |
| "learning_rate": 1.4808582075166778e-06, | |
| "loss": 0.2487, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25047406554222107, | |
| "step": 3480, | |
| "valid_targets_mean": 4213.7, | |
| "valid_targets_min": 1750 | |
| }, | |
| { | |
| "epoch": 6.2343470483005365, | |
| "grad_norm": 0.594758118058126, | |
| "learning_rate": 1.447348893118965e-06, | |
| "loss": 0.2446, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2589150667190552, | |
| "step": 3485, | |
| "valid_targets_mean": 3918.9, | |
| "valid_targets_min": 2197 | |
| }, | |
| { | |
| "epoch": 6.243291592128801, | |
| "grad_norm": 0.5390376277621793, | |
| "learning_rate": 1.4142088223081408e-06, | |
| "loss": 0.2441, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24606183171272278, | |
| "step": 3490, | |
| "valid_targets_mean": 4628.6, | |
| "valid_targets_min": 2230 | |
| }, | |
| { | |
| "epoch": 6.252236135957066, | |
| "grad_norm": 0.515173541243895, | |
| "learning_rate": 1.3814386546534864e-06, | |
| "loss": 0.253, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2567700147628784, | |
| "step": 3495, | |
| "valid_targets_mean": 4171.9, | |
| "valid_targets_min": 1779 | |
| }, | |
| { | |
| "epoch": 6.261180679785331, | |
| "grad_norm": 0.5308974195526618, | |
| "learning_rate": 1.3490390423622856e-06, | |
| "loss": 0.2427, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22043094038963318, | |
| "step": 3500, | |
| "valid_targets_mean": 4421.9, | |
| "valid_targets_min": 1412 | |
| }, | |
| { | |
| "epoch": 6.270125223613595, | |
| "grad_norm": 0.5045875631623113, | |
| "learning_rate": 1.3170106302668616e-06, | |
| "loss": 0.2455, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23024529218673706, | |
| "step": 3505, | |
| "valid_targets_mean": 4453.3, | |
| "valid_targets_min": 1660 | |
| }, | |
| { | |
| "epoch": 6.27906976744186, | |
| "grad_norm": 0.5448476578553636, | |
| "learning_rate": 1.2853540558117272e-06, | |
| "loss": 0.2454, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24999341368675232, | |
| "step": 3510, | |
| "valid_targets_mean": 4241.5, | |
| "valid_targets_min": 1712 | |
| }, | |
| { | |
| "epoch": 6.288014311270125, | |
| "grad_norm": 0.5759478717126266, | |
| "learning_rate": 1.254069949040917e-06, | |
| "loss": 0.254, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25016283988952637, | |
| "step": 3515, | |
| "valid_targets_mean": 3847.2, | |
| "valid_targets_min": 669 | |
| }, | |
| { | |
| "epoch": 6.29695885509839, | |
| "grad_norm": 0.542415226047018, | |
| "learning_rate": 1.2231589325854354e-06, | |
| "loss": 0.2474, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2293107956647873, | |
| "step": 3520, | |
| "valid_targets_mean": 4337.9, | |
| "valid_targets_min": 1502 | |
| }, | |
| { | |
| "epoch": 6.305903398926655, | |
| "grad_norm": 0.5091719583649115, | |
| "learning_rate": 1.1926216216508735e-06, | |
| "loss": 0.2565, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25717687606811523, | |
| "step": 3525, | |
| "valid_targets_mean": 4432.9, | |
| "valid_targets_min": 737 | |
| }, | |
| { | |
| "epoch": 6.31484794275492, | |
| "grad_norm": 0.4941207542833962, | |
| "learning_rate": 1.162458624005145e-06, | |
| "loss": 0.2525, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24258843064308167, | |
| "step": 3530, | |
| "valid_targets_mean": 4559.1, | |
| "valid_targets_min": 1824 | |
| }, | |
| { | |
| "epoch": 6.3237924865831845, | |
| "grad_norm": 0.5967014559288393, | |
| "learning_rate": 1.1326705399664207e-06, | |
| "loss": 0.2451, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23474426567554474, | |
| "step": 3535, | |
| "valid_targets_mean": 3250.2, | |
| "valid_targets_min": 1600 | |
| }, | |
| { | |
| "epoch": 6.332737030411449, | |
| "grad_norm": 0.5207543513220922, | |
| "learning_rate": 1.103257962391151e-06, | |
| "loss": 0.237, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24193906784057617, | |
| "step": 3540, | |
| "valid_targets_mean": 4304.2, | |
| "valid_targets_min": 763 | |
| }, | |
| { | |
| "epoch": 6.341681574239714, | |
| "grad_norm": 0.5920796448944554, | |
| "learning_rate": 1.0742214766622938e-06, | |
| "loss": 0.2523, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2564089596271515, | |
| "step": 3545, | |
| "valid_targets_mean": 3882.9, | |
| "valid_targets_min": 579 | |
| }, | |
| { | |
| "epoch": 6.350626118067979, | |
| "grad_norm": 0.5232360068927403, | |
| "learning_rate": 1.0455616606776363e-06, | |
| "loss": 0.2395, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2539476156234741, | |
| "step": 3550, | |
| "valid_targets_mean": 3925.2, | |
| "valid_targets_min": 805 | |
| }, | |
| { | |
| "epoch": 6.359570661896243, | |
| "grad_norm": 0.4866461666276335, | |
| "learning_rate": 1.0172790848383141e-06, | |
| "loss": 0.2435, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24338996410369873, | |
| "step": 3555, | |
| "valid_targets_mean": 4900.0, | |
| "valid_targets_min": 2526 | |
| }, | |
| { | |
| "epoch": 6.368515205724508, | |
| "grad_norm": 0.5299824855199575, | |
| "learning_rate": 9.893743120374543e-07, | |
| "loss": 0.2465, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24363267421722412, | |
| "step": 3560, | |
| "valid_targets_mean": 3905.7, | |
| "valid_targets_min": 1083 | |
| }, | |
| { | |
| "epoch": 6.377459749552773, | |
| "grad_norm": 0.5642560950283441, | |
| "learning_rate": 9.618478976489709e-07, | |
| "loss": 0.2517, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23905110359191895, | |
| "step": 3565, | |
| "valid_targets_mean": 4138.6, | |
| "valid_targets_min": 856 | |
| }, | |
| { | |
| "epoch": 6.386404293381037, | |
| "grad_norm": 0.5115018928118804, | |
| "learning_rate": 9.347003895165052e-07, | |
| "loss": 0.2448, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26956266164779663, | |
| "step": 3570, | |
| "valid_targets_mean": 5040.1, | |
| "valid_targets_min": 1505 | |
| }, | |
| { | |
| "epoch": 6.395348837209302, | |
| "grad_norm": 0.5710236424221047, | |
| "learning_rate": 9.079323279425245e-07, | |
| "loss": 0.2532, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2561646103858948, | |
| "step": 3575, | |
| "valid_targets_mean": 4197.1, | |
| "valid_targets_min": 1158 | |
| }, | |
| { | |
| "epoch": 6.404293381037567, | |
| "grad_norm": 0.5054247186866906, | |
| "learning_rate": 8.815442456775835e-07, | |
| "loss": 0.249, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25438791513442993, | |
| "step": 3580, | |
| "valid_targets_mean": 4520.4, | |
| "valid_targets_min": 1430 | |
| }, | |
| { | |
| "epoch": 6.4132379248658316, | |
| "grad_norm": 0.5325941678660696, | |
| "learning_rate": 8.555366679097043e-07, | |
| "loss": 0.2508, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24359354376792908, | |
| "step": 3585, | |
| "valid_targets_mean": 4353.4, | |
| "valid_targets_min": 1339 | |
| }, | |
| { | |
| "epoch": 6.422182468694096, | |
| "grad_norm": 0.5189637715129809, | |
| "learning_rate": 8.29910112253931e-07, | |
| "loss": 0.2492, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2488621473312378, | |
| "step": 3590, | |
| "valid_targets_mean": 4283.8, | |
| "valid_targets_min": 2266 | |
| }, | |
| { | |
| "epoch": 6.431127012522361, | |
| "grad_norm": 0.5866424542601351, | |
| "learning_rate": 8.046650887420227e-07, | |
| "loss": 0.2456, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26135364174842834, | |
| "step": 3595, | |
| "valid_targets_mean": 3206.2, | |
| "valid_targets_min": 1617 | |
| }, | |
| { | |
| "epoch": 6.440071556350626, | |
| "grad_norm": 0.5499318781593886, | |
| "learning_rate": 7.798020998123101e-07, | |
| "loss": 0.2471, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2563382685184479, | |
| "step": 3600, | |
| "valid_targets_mean": 4393.6, | |
| "valid_targets_min": 710 | |
| }, | |
| { | |
| "epoch": 6.449016100178891, | |
| "grad_norm": 0.5310211310663512, | |
| "learning_rate": 7.55321640299691e-07, | |
| "loss": 0.248, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25477510690689087, | |
| "step": 3605, | |
| "valid_targets_mean": 4263.4, | |
| "valid_targets_min": 2011 | |
| }, | |
| { | |
| "epoch": 6.457960644007156, | |
| "grad_norm": 0.560059078942515, | |
| "learning_rate": 7.312241974257861e-07, | |
| "loss": 0.2373, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2293008267879486, | |
| "step": 3610, | |
| "valid_targets_mean": 3617.6, | |
| "valid_targets_min": 1590 | |
| }, | |
| { | |
| "epoch": 6.466905187835421, | |
| "grad_norm": 0.5267478780824102, | |
| "learning_rate": 7.075102507892295e-07, | |
| "loss": 0.247, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23703113198280334, | |
| "step": 3615, | |
| "valid_targets_mean": 4467.4, | |
| "valid_targets_min": 1654 | |
| }, | |
| { | |
| "epoch": 6.475849731663685, | |
| "grad_norm": 0.5537266130087152, | |
| "learning_rate": 6.841802723561408e-07, | |
| "loss": 0.2325, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.241033136844635, | |
| "step": 3620, | |
| "valid_targets_mean": 4434.5, | |
| "valid_targets_min": 680 | |
| }, | |
| { | |
| "epoch": 6.48479427549195, | |
| "grad_norm": 0.5258675827998067, | |
| "learning_rate": 6.612347264507257e-07, | |
| "loss": 0.2569, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23057600855827332, | |
| "step": 3625, | |
| "valid_targets_mean": 4508.6, | |
| "valid_targets_min": 1675 | |
| }, | |
| { | |
| "epoch": 6.493738819320215, | |
| "grad_norm": 0.6060163526427282, | |
| "learning_rate": 6.386740697460281e-07, | |
| "loss": 0.2582, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.266754686832428, | |
| "step": 3630, | |
| "valid_targets_mean": 3405.2, | |
| "valid_targets_min": 1494 | |
| }, | |
| { | |
| "epoch": 6.5026833631484795, | |
| "grad_norm": 1.243756853379082, | |
| "learning_rate": 6.164987512548415e-07, | |
| "loss": 0.2425, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24251213669776917, | |
| "step": 3635, | |
| "valid_targets_mean": 4815.0, | |
| "valid_targets_min": 748 | |
| }, | |
| { | |
| "epoch": 6.511627906976744, | |
| "grad_norm": 0.4905512612650833, | |
| "learning_rate": 5.947092123207854e-07, | |
| "loss": 0.2458, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2418278455734253, | |
| "step": 3640, | |
| "valid_targets_mean": 4851.4, | |
| "valid_targets_min": 1723 | |
| }, | |
| { | |
| "epoch": 6.520572450805009, | |
| "grad_norm": 0.6977038416301831, | |
| "learning_rate": 5.733058866095143e-07, | |
| "loss": 0.2439, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25958025455474854, | |
| "step": 3645, | |
| "valid_targets_mean": 3321.6, | |
| "valid_targets_min": 864 | |
| }, | |
| { | |
| "epoch": 6.529516994633274, | |
| "grad_norm": 0.5620237411481044, | |
| "learning_rate": 5.522892001000801e-07, | |
| "loss": 0.247, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25899815559387207, | |
| "step": 3650, | |
| "valid_targets_mean": 3941.6, | |
| "valid_targets_min": 645 | |
| }, | |
| { | |
| "epoch": 6.538461538461538, | |
| "grad_norm": 0.5931969111624182, | |
| "learning_rate": 5.316595710764594e-07, | |
| "loss": 0.2485, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2590034008026123, | |
| "step": 3655, | |
| "valid_targets_mean": 3292.1, | |
| "valid_targets_min": 1140 | |
| }, | |
| { | |
| "epoch": 6.547406082289803, | |
| "grad_norm": 0.5619298957004911, | |
| "learning_rate": 5.114174101192282e-07, | |
| "loss": 0.2551, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25470370054244995, | |
| "step": 3660, | |
| "valid_targets_mean": 4170.2, | |
| "valid_targets_min": 1529 | |
| }, | |
| { | |
| "epoch": 6.556350626118068, | |
| "grad_norm": 0.5338888004490264, | |
| "learning_rate": 4.915631200973958e-07, | |
| "loss": 0.2489, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23330549895763397, | |
| "step": 3665, | |
| "valid_targets_mean": 4152.4, | |
| "valid_targets_min": 1548 | |
| }, | |
| { | |
| "epoch": 6.565295169946332, | |
| "grad_norm": 0.5029110114373035, | |
| "learning_rate": 4.7209709616038214e-07, | |
| "loss": 0.2413, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2532098591327667, | |
| "step": 3670, | |
| "valid_targets_mean": 4623.9, | |
| "valid_targets_min": 1504 | |
| }, | |
| { | |
| "epoch": 6.574239713774597, | |
| "grad_norm": 0.5553623774184658, | |
| "learning_rate": 4.5301972573014163e-07, | |
| "loss": 0.2498, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25203073024749756, | |
| "step": 3675, | |
| "valid_targets_mean": 4177.2, | |
| "valid_targets_min": 954 | |
| }, | |
| { | |
| "epoch": 6.583184257602863, | |
| "grad_norm": 0.5645036118540637, | |
| "learning_rate": 4.3433138849347854e-07, | |
| "loss": 0.255, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25823813676834106, | |
| "step": 3680, | |
| "valid_targets_mean": 4475.9, | |
| "valid_targets_min": 2315 | |
| }, | |
| { | |
| "epoch": 6.592128801431127, | |
| "grad_norm": 0.5703760997362556, | |
| "learning_rate": 4.1603245639446845e-07, | |
| "loss": 0.2475, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24437285959720612, | |
| "step": 3685, | |
| "valid_targets_mean": 4044.4, | |
| "valid_targets_min": 1603 | |
| }, | |
| { | |
| "epoch": 6.601073345259392, | |
| "grad_norm": 0.5422079585581118, | |
| "learning_rate": 3.981232936270596e-07, | |
| "loss": 0.2324, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23096385598182678, | |
| "step": 3690, | |
| "valid_targets_mean": 4084.5, | |
| "valid_targets_min": 2212 | |
| }, | |
| { | |
| "epoch": 6.610017889087657, | |
| "grad_norm": 0.560268180958077, | |
| "learning_rate": 3.806042566278345e-07, | |
| "loss": 0.2623, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3074514865875244, | |
| "step": 3695, | |
| "valid_targets_mean": 4345.9, | |
| "valid_targets_min": 1526 | |
| }, | |
| { | |
| "epoch": 6.618962432915922, | |
| "grad_norm": 0.5139061022298529, | |
| "learning_rate": 3.6347569406890213e-07, | |
| "loss": 0.2439, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24795208871364594, | |
| "step": 3700, | |
| "valid_targets_mean": 4453.6, | |
| "valid_targets_min": 2173 | |
| }, | |
| { | |
| "epoch": 6.627906976744186, | |
| "grad_norm": 0.5732304901299363, | |
| "learning_rate": 3.467379468509746e-07, | |
| "loss": 0.2512, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2507464289665222, | |
| "step": 3705, | |
| "valid_targets_mean": 3565.4, | |
| "valid_targets_min": 1726 | |
| }, | |
| { | |
| "epoch": 6.636851520572451, | |
| "grad_norm": 0.5275323599021355, | |
| "learning_rate": 3.3039134809656594e-07, | |
| "loss": 0.238, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25029295682907104, | |
| "step": 3710, | |
| "valid_targets_mean": 4312.4, | |
| "valid_targets_min": 1826 | |
| }, | |
| { | |
| "epoch": 6.645796064400716, | |
| "grad_norm": 0.5142402557074688, | |
| "learning_rate": 3.144362231433706e-07, | |
| "loss": 0.2442, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24940407276153564, | |
| "step": 3715, | |
| "valid_targets_mean": 4422.9, | |
| "valid_targets_min": 747 | |
| }, | |
| { | |
| "epoch": 6.65474060822898, | |
| "grad_norm": 0.5730417596248877, | |
| "learning_rate": 2.9887288953778903e-07, | |
| "loss": 0.2564, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2630217671394348, | |
| "step": 3720, | |
| "valid_targets_mean": 4032.0, | |
| "valid_targets_min": 1372 | |
| }, | |
| { | |
| "epoch": 6.663685152057245, | |
| "grad_norm": 0.554437674633871, | |
| "learning_rate": 2.8370165702860553e-07, | |
| "loss": 0.2451, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26767322421073914, | |
| "step": 3725, | |
| "valid_targets_mean": 4060.4, | |
| "valid_targets_min": 1319 | |
| }, | |
| { | |
| "epoch": 6.67262969588551, | |
| "grad_norm": 0.5716086807724999, | |
| "learning_rate": 2.6892282756082246e-07, | |
| "loss": 0.246, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23963013291358948, | |
| "step": 3730, | |
| "valid_targets_mean": 3399.0, | |
| "valid_targets_min": 691 | |
| }, | |
| { | |
| "epoch": 6.6815742397137745, | |
| "grad_norm": 0.5079641007584742, | |
| "learning_rate": 2.545366952696582e-07, | |
| "loss": 0.2441, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2464374452829361, | |
| "step": 3735, | |
| "valid_targets_mean": 4499.1, | |
| "valid_targets_min": 1971 | |
| }, | |
| { | |
| "epoch": 6.690518783542039, | |
| "grad_norm": 0.9402640194261683, | |
| "learning_rate": 2.405435464746786e-07, | |
| "loss": 0.264, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27681803703308105, | |
| "step": 3740, | |
| "valid_targets_mean": 3995.8, | |
| "valid_targets_min": 656 | |
| }, | |
| { | |
| "epoch": 6.699463327370304, | |
| "grad_norm": 0.5895147291124684, | |
| "learning_rate": 2.2694365967411258e-07, | |
| "loss": 0.2572, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26021093130111694, | |
| "step": 3745, | |
| "valid_targets_mean": 3512.1, | |
| "valid_targets_min": 922 | |
| }, | |
| { | |
| "epoch": 6.708407871198569, | |
| "grad_norm": 0.5247512021335333, | |
| "learning_rate": 2.137373055393055e-07, | |
| "loss": 0.2525, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2672116756439209, | |
| "step": 3750, | |
| "valid_targets_mean": 4461.2, | |
| "valid_targets_min": 1031 | |
| }, | |
| { | |
| "epoch": 6.717352415026833, | |
| "grad_norm": 0.5648115569047794, | |
| "learning_rate": 2.009247469093234e-07, | |
| "loss": 0.2372, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24653485417366028, | |
| "step": 3755, | |
| "valid_targets_mean": 3891.8, | |
| "valid_targets_min": 1329 | |
| }, | |
| { | |
| "epoch": 6.726296958855098, | |
| "grad_norm": 0.5935904883224387, | |
| "learning_rate": 1.8850623878573505e-07, | |
| "loss": 0.2541, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2428642213344574, | |
| "step": 3760, | |
| "valid_targets_mean": 3706.3, | |
| "valid_targets_min": 931 | |
| }, | |
| { | |
| "epoch": 6.735241502683364, | |
| "grad_norm": 0.5282006716958451, | |
| "learning_rate": 1.7648202832752924e-07, | |
| "loss": 0.2583, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22287273406982422, | |
| "step": 3765, | |
| "valid_targets_mean": 4499.3, | |
| "valid_targets_min": 1662 | |
| }, | |
| { | |
| "epoch": 6.7441860465116275, | |
| "grad_norm": 0.5697571140348834, | |
| "learning_rate": 1.6485235484619665e-07, | |
| "loss": 0.2428, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25361984968185425, | |
| "step": 3770, | |
| "valid_targets_mean": 3464.8, | |
| "valid_targets_min": 1843 | |
| }, | |
| { | |
| "epoch": 6.753130590339893, | |
| "grad_norm": 0.561107497561089, | |
| "learning_rate": 1.5361744980096905e-07, | |
| "loss": 0.2382, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23007211089134216, | |
| "step": 3775, | |
| "valid_targets_mean": 3861.9, | |
| "valid_targets_min": 1742 | |
| }, | |
| { | |
| "epoch": 6.762075134168158, | |
| "grad_norm": 0.5240522175257502, | |
| "learning_rate": 1.427775367942097e-07, | |
| "loss": 0.2564, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25991159677505493, | |
| "step": 3780, | |
| "valid_targets_mean": 4519.6, | |
| "valid_targets_min": 1936 | |
| }, | |
| { | |
| "epoch": 6.7710196779964225, | |
| "grad_norm": 0.5553928938449468, | |
| "learning_rate": 1.3233283156696142e-07, | |
| "loss": 0.2437, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2548370659351349, | |
| "step": 3785, | |
| "valid_targets_mean": 3997.4, | |
| "valid_targets_min": 621 | |
| }, | |
| { | |
| "epoch": 6.779964221824687, | |
| "grad_norm": 0.5074568016914366, | |
| "learning_rate": 1.222835419946633e-07, | |
| "loss": 0.252, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23944196105003357, | |
| "step": 3790, | |
| "valid_targets_mean": 5238.7, | |
| "valid_targets_min": 2620 | |
| }, | |
| { | |
| "epoch": 6.788908765652952, | |
| "grad_norm": 0.5333184632049403, | |
| "learning_rate": 1.1262986808300291e-07, | |
| "loss": 0.2485, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25998786091804504, | |
| "step": 3795, | |
| "valid_targets_mean": 4385.2, | |
| "valid_targets_min": 1617 | |
| }, | |
| { | |
| "epoch": 6.797853309481217, | |
| "grad_norm": 0.5068220624841174, | |
| "learning_rate": 1.0337200196393505e-07, | |
| "loss": 0.2577, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25370797514915466, | |
| "step": 3800, | |
| "valid_targets_mean": 4350.8, | |
| "valid_targets_min": 1473 | |
| }, | |
| { | |
| "epoch": 6.806797853309481, | |
| "grad_norm": 0.5340749980997674, | |
| "learning_rate": 9.451012789186476e-08, | |
| "loss": 0.2433, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24803949892520905, | |
| "step": 3805, | |
| "valid_targets_mean": 4403.3, | |
| "valid_targets_min": 784 | |
| }, | |
| { | |
| "epoch": 6.815742397137746, | |
| "grad_norm": 0.5465845662554875, | |
| "learning_rate": 8.604442223998144e-08, | |
| "loss": 0.2621, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28449052572250366, | |
| "step": 3810, | |
| "valid_targets_mean": 4181.7, | |
| "valid_targets_min": 1019 | |
| }, | |
| { | |
| "epoch": 6.824686940966011, | |
| "grad_norm": 0.5740109665058152, | |
| "learning_rate": 7.797505349673496e-08, | |
| "loss": 0.2428, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24289534986019135, | |
| "step": 3815, | |
| "valid_targets_mean": 3507.5, | |
| "valid_targets_min": 1523 | |
| }, | |
| { | |
| "epoch": 6.833631484794275, | |
| "grad_norm": 0.532987766803314, | |
| "learning_rate": 7.030218226250051e-08, | |
| "loss": 0.251, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24493467807769775, | |
| "step": 3820, | |
| "valid_targets_mean": 4069.8, | |
| "valid_targets_min": 795 | |
| }, | |
| { | |
| "epoch": 6.84257602862254, | |
| "grad_norm": 0.6131271407710378, | |
| "learning_rate": 6.302596124636573e-08, | |
| "loss": 0.2555, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2517031729221344, | |
| "step": 3825, | |
| "valid_targets_mean": 4256.1, | |
| "valid_targets_min": 741 | |
| }, | |
| { | |
| "epoch": 6.851520572450805, | |
| "grad_norm": 0.5260881026648205, | |
| "learning_rate": 5.6146535263106314e-08, | |
| "loss": 0.249, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22869578003883362, | |
| "step": 3830, | |
| "valid_targets_mean": 4011.5, | |
| "valid_targets_min": 1329 | |
| }, | |
| { | |
| "epoch": 6.8604651162790695, | |
| "grad_norm": 0.6127050009448671, | |
| "learning_rate": 4.966404123029289e-08, | |
| "loss": 0.2448, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2303614318370819, | |
| "step": 3835, | |
| "valid_targets_mean": 4033.9, | |
| "valid_targets_min": 1573 | |
| }, | |
| { | |
| "epoch": 6.869409660107334, | |
| "grad_norm": 0.5818917098959172, | |
| "learning_rate": 4.357860816557091e-08, | |
| "loss": 0.2611, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2567775249481201, | |
| "step": 3840, | |
| "valid_targets_mean": 3792.1, | |
| "valid_targets_min": 529 | |
| }, | |
| { | |
| "epoch": 6.878354203935599, | |
| "grad_norm": 0.5360181923150845, | |
| "learning_rate": 3.789035718408718e-08, | |
| "loss": 0.2458, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24667097628116608, | |
| "step": 3845, | |
| "valid_targets_mean": 3941.6, | |
| "valid_targets_min": 1016 | |
| }, | |
| { | |
| "epoch": 6.8872987477638645, | |
| "grad_norm": 0.596035881473606, | |
| "learning_rate": 3.259940149609175e-08, | |
| "loss": 0.2522, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.278276652097702, | |
| "step": 3850, | |
| "valid_targets_mean": 3848.6, | |
| "valid_targets_min": 884 | |
| }, | |
| { | |
| "epoch": 6.896243291592128, | |
| "grad_norm": 0.5426279598087133, | |
| "learning_rate": 2.7705846404670888e-08, | |
| "loss": 0.2362, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25362810492515564, | |
| "step": 3855, | |
| "valid_targets_mean": 3896.8, | |
| "valid_targets_min": 1947 | |
| }, | |
| { | |
| "epoch": 6.905187835420394, | |
| "grad_norm": 0.5217040663626972, | |
| "learning_rate": 2.3209789303657582e-08, | |
| "loss": 0.2428, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24874532222747803, | |
| "step": 3860, | |
| "valid_targets_mean": 4455.1, | |
| "valid_targets_min": 1374 | |
| }, | |
| { | |
| "epoch": 6.914132379248659, | |
| "grad_norm": 0.6253844624269363, | |
| "learning_rate": 1.9111319675699792e-08, | |
| "loss": 0.2447, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24061639606952667, | |
| "step": 3865, | |
| "valid_targets_mean": 4007.6, | |
| "valid_targets_min": 1409 | |
| }, | |
| { | |
| "epoch": 6.923076923076923, | |
| "grad_norm": 0.600953932839311, | |
| "learning_rate": 1.5410519090461872e-08, | |
| "loss": 0.2478, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25711876153945923, | |
| "step": 3870, | |
| "valid_targets_mean": 3362.2, | |
| "valid_targets_min": 995 | |
| }, | |
| { | |
| "epoch": 6.932021466905188, | |
| "grad_norm": 0.5660567298832868, | |
| "learning_rate": 1.210746120302142e-08, | |
| "loss": 0.2463, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2501237392425537, | |
| "step": 3875, | |
| "valid_targets_mean": 3972.0, | |
| "valid_targets_min": 2149 | |
| }, | |
| { | |
| "epoch": 6.940966010733453, | |
| "grad_norm": 0.6014456216364971, | |
| "learning_rate": 9.202211752386003e-09, | |
| "loss": 0.2541, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25210267305374146, | |
| "step": 3880, | |
| "valid_targets_mean": 4391.4, | |
| "valid_targets_min": 1615 | |
| }, | |
| { | |
| "epoch": 6.9499105545617175, | |
| "grad_norm": 0.5988847243290031, | |
| "learning_rate": 6.694828560200872e-09, | |
| "loss": 0.2572, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2896597385406494, | |
| "step": 3885, | |
| "valid_targets_mean": 4030.7, | |
| "valid_targets_min": 1853 | |
| }, | |
| { | |
| "epoch": 6.958855098389982, | |
| "grad_norm": 0.5395682499202165, | |
| "learning_rate": 4.585361529581001e-09, | |
| "loss": 0.2411, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2411486953496933, | |
| "step": 3890, | |
| "valid_targets_mean": 4078.4, | |
| "valid_targets_min": 822 | |
| }, | |
| { | |
| "epoch": 6.967799642218247, | |
| "grad_norm": 0.6335748048992399, | |
| "learning_rate": 2.87385264413631e-09, | |
| "loss": 0.2559, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2603016793727875, | |
| "step": 3895, | |
| "valid_targets_mean": 2954.7, | |
| "valid_targets_min": 911 | |
| }, | |
| { | |
| "epoch": 6.976744186046512, | |
| "grad_norm": 0.5727655803906435, | |
| "learning_rate": 1.560335967114579e-09, | |
| "loss": 0.2541, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25090134143829346, | |
| "step": 3900, | |
| "valid_targets_mean": 4419.8, | |
| "valid_targets_min": 1517 | |
| }, | |
| { | |
| "epoch": 6.985688729874776, | |
| "grad_norm": 0.5157971974589199, | |
| "learning_rate": 6.4483764074863e-10, | |
| "loss": 0.2364, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2580246329307556, | |
| "step": 3905, | |
| "valid_targets_mean": 4267.4, | |
| "valid_targets_min": 1483 | |
| }, | |
| { | |
| "epoch": 6.994633273703041, | |
| "grad_norm": 0.5412701822718862, | |
| "learning_rate": 1.273758857145424e-10, | |
| "loss": 0.2505, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2360781729221344, | |
| "step": 3910, | |
| "valid_targets_mean": 3865.5, | |
| "valid_targets_min": 653 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23571906983852386, | |
| "step": 3913, | |
| "total_flos": 1325319151616000.0, | |
| "train_loss": 0.29724271997234386, | |
| "train_runtime": 27386.6627, | |
| "train_samples_per_second": 2.285, | |
| "train_steps_per_second": 0.143, | |
| "valid_targets_mean": 3601.1, | |
| "valid_targets_min": 1303 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 3913, | |
| "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": 1325319151616000.0, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |