Text Generation
Transformers
Safetensors
qwen3
llama-factory
full
Generated from Trainer
conversational
text-generation-inference
Instructions to use DCAgent/a1-exercism_python with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use DCAgent/a1-exercism_python with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="DCAgent/a1-exercism_python") messages = [ {"role": "user", "content": "Who are you?"}, ] pipe(messages)# Load model directly from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("DCAgent/a1-exercism_python") model = AutoModelForCausalLM.from_pretrained("DCAgent/a1-exercism_python", 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-exercism_python with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "DCAgent/a1-exercism_python" # 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-exercism_python", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/DCAgent/a1-exercism_python
- SGLang
How to use DCAgent/a1-exercism_python 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-exercism_python" \ --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-exercism_python", "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-exercism_python" \ --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-exercism_python", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }' - Docker Model Runner
How to use DCAgent/a1-exercism_python with Docker Model Runner:
docker model run hf.co/DCAgent/a1-exercism_python
| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 7.0, | |
| "eval_steps": 500, | |
| "global_step": 4389, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.007974481658692184, | |
| "grad_norm": 19.08310269585043, | |
| "learning_rate": 3.644646924829157e-07, | |
| "loss": 0.7989, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.8064001798629761, | |
| "step": 5, | |
| "valid_targets_mean": 3405.2, | |
| "valid_targets_min": 1804 | |
| }, | |
| { | |
| "epoch": 0.01594896331738437, | |
| "grad_norm": 18.15244545055453, | |
| "learning_rate": 8.200455580865605e-07, | |
| "loss": 0.7787, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.7641298174858093, | |
| "step": 10, | |
| "valid_targets_mean": 2772.0, | |
| "valid_targets_min": 1037 | |
| }, | |
| { | |
| "epoch": 0.023923444976076555, | |
| "grad_norm": 13.672484690698322, | |
| "learning_rate": 1.2756264236902052e-06, | |
| "loss": 0.7357, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.6846181154251099, | |
| "step": 15, | |
| "valid_targets_mean": 3543.5, | |
| "valid_targets_min": 1320 | |
| }, | |
| { | |
| "epoch": 0.03189792663476874, | |
| "grad_norm": 12.741705912118789, | |
| "learning_rate": 1.7312072892938498e-06, | |
| "loss": 0.7002, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.6495967507362366, | |
| "step": 20, | |
| "valid_targets_mean": 3172.8, | |
| "valid_targets_min": 1190 | |
| }, | |
| { | |
| "epoch": 0.03987240829346093, | |
| "grad_norm": 8.796778470910523, | |
| "learning_rate": 2.1867881548974945e-06, | |
| "loss": 0.6617, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.5894312262535095, | |
| "step": 25, | |
| "valid_targets_mean": 3213.4, | |
| "valid_targets_min": 1369 | |
| }, | |
| { | |
| "epoch": 0.04784688995215311, | |
| "grad_norm": 5.349177988575253, | |
| "learning_rate": 2.642369020501139e-06, | |
| "loss": 0.6311, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.6095372438430786, | |
| "step": 30, | |
| "valid_targets_mean": 3185.8, | |
| "valid_targets_min": 1198 | |
| }, | |
| { | |
| "epoch": 0.05582137161084529, | |
| "grad_norm": 3.3314893539168238, | |
| "learning_rate": 3.0979498861047843e-06, | |
| "loss": 0.5367, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.4944782853126526, | |
| "step": 35, | |
| "valid_targets_mean": 2579.5, | |
| "valid_targets_min": 376 | |
| }, | |
| { | |
| "epoch": 0.06379585326953748, | |
| "grad_norm": 2.4005782583446984, | |
| "learning_rate": 3.5535307517084285e-06, | |
| "loss": 0.5448, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.5224937200546265, | |
| "step": 40, | |
| "valid_targets_mean": 2636.4, | |
| "valid_targets_min": 1297 | |
| }, | |
| { | |
| "epoch": 0.07177033492822966, | |
| "grad_norm": 1.3723814844051554, | |
| "learning_rate": 4.009111617312073e-06, | |
| "loss": 0.478, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.4328991770744324, | |
| "step": 45, | |
| "valid_targets_mean": 3519.6, | |
| "valid_targets_min": 1141 | |
| }, | |
| { | |
| "epoch": 0.07974481658692185, | |
| "grad_norm": 1.3026590573106005, | |
| "learning_rate": 4.464692482915718e-06, | |
| "loss": 0.5271, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.4638175070285797, | |
| "step": 50, | |
| "valid_targets_mean": 2428.6, | |
| "valid_targets_min": 1075 | |
| }, | |
| { | |
| "epoch": 0.08771929824561403, | |
| "grad_norm": 1.2999173286453858, | |
| "learning_rate": 4.920273348519363e-06, | |
| "loss": 0.4904, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.5091861486434937, | |
| "step": 55, | |
| "valid_targets_mean": 2451.5, | |
| "valid_targets_min": 1214 | |
| }, | |
| { | |
| "epoch": 0.09569377990430622, | |
| "grad_norm": 0.911171533209259, | |
| "learning_rate": 5.375854214123008e-06, | |
| "loss": 0.4631, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.4018824100494385, | |
| "step": 60, | |
| "valid_targets_mean": 2899.0, | |
| "valid_targets_min": 1181 | |
| }, | |
| { | |
| "epoch": 0.10366826156299841, | |
| "grad_norm": 0.9137218681759303, | |
| "learning_rate": 5.831435079726651e-06, | |
| "loss": 0.4235, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.4257986545562744, | |
| "step": 65, | |
| "valid_targets_mean": 2692.9, | |
| "valid_targets_min": 1217 | |
| }, | |
| { | |
| "epoch": 0.11164274322169059, | |
| "grad_norm": 0.7940217799338982, | |
| "learning_rate": 6.287015945330297e-06, | |
| "loss": 0.3964, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3876068592071533, | |
| "step": 70, | |
| "valid_targets_mean": 2944.8, | |
| "valid_targets_min": 986 | |
| }, | |
| { | |
| "epoch": 0.11961722488038277, | |
| "grad_norm": 0.8646607237411834, | |
| "learning_rate": 6.742596810933942e-06, | |
| "loss": 0.4214, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3561863303184509, | |
| "step": 75, | |
| "valid_targets_mean": 2176.2, | |
| "valid_targets_min": 1028 | |
| }, | |
| { | |
| "epoch": 0.12759170653907495, | |
| "grad_norm": 0.7019349077333937, | |
| "learning_rate": 7.1981776765375854e-06, | |
| "loss": 0.3958, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.36579617857933044, | |
| "step": 80, | |
| "valid_targets_mean": 3109.9, | |
| "valid_targets_min": 1427 | |
| }, | |
| { | |
| "epoch": 0.13556618819776714, | |
| "grad_norm": 0.776844100371323, | |
| "learning_rate": 7.65375854214123e-06, | |
| "loss": 0.4248, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3523101508617401, | |
| "step": 85, | |
| "valid_targets_mean": 2283.4, | |
| "valid_targets_min": 1124 | |
| }, | |
| { | |
| "epoch": 0.14354066985645933, | |
| "grad_norm": 0.7814197273845711, | |
| "learning_rate": 8.109339407744875e-06, | |
| "loss": 0.3677, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.36872315406799316, | |
| "step": 90, | |
| "valid_targets_mean": 2170.3, | |
| "valid_targets_min": 1321 | |
| }, | |
| { | |
| "epoch": 0.15151515151515152, | |
| "grad_norm": 0.6933045488358227, | |
| "learning_rate": 8.564920273348521e-06, | |
| "loss": 0.3791, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3503817319869995, | |
| "step": 95, | |
| "valid_targets_mean": 2544.8, | |
| "valid_targets_min": 1054 | |
| }, | |
| { | |
| "epoch": 0.1594896331738437, | |
| "grad_norm": 0.7344042311376202, | |
| "learning_rate": 9.020501138952164e-06, | |
| "loss": 0.4041, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.36823010444641113, | |
| "step": 100, | |
| "valid_targets_mean": 2677.9, | |
| "valid_targets_min": 1397 | |
| }, | |
| { | |
| "epoch": 0.1674641148325359, | |
| "grad_norm": 0.8885061382631938, | |
| "learning_rate": 9.47608200455581e-06, | |
| "loss": 0.3574, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.37888795137405396, | |
| "step": 105, | |
| "valid_targets_mean": 2963.9, | |
| "valid_targets_min": 1098 | |
| }, | |
| { | |
| "epoch": 0.17543859649122806, | |
| "grad_norm": 0.7040180595260014, | |
| "learning_rate": 9.931662870159453e-06, | |
| "loss": 0.3907, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.4308871626853943, | |
| "step": 110, | |
| "valid_targets_mean": 2947.4, | |
| "valid_targets_min": 1378 | |
| }, | |
| { | |
| "epoch": 0.18341307814992025, | |
| "grad_norm": 0.6704086894938297, | |
| "learning_rate": 1.03872437357631e-05, | |
| "loss": 0.3819, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.374492347240448, | |
| "step": 115, | |
| "valid_targets_mean": 2938.1, | |
| "valid_targets_min": 528 | |
| }, | |
| { | |
| "epoch": 0.19138755980861244, | |
| "grad_norm": 0.7460794224492051, | |
| "learning_rate": 1.0842824601366744e-05, | |
| "loss": 0.3688, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3808710277080536, | |
| "step": 120, | |
| "valid_targets_mean": 2539.7, | |
| "valid_targets_min": 862 | |
| }, | |
| { | |
| "epoch": 0.19936204146730463, | |
| "grad_norm": 0.7530056967637131, | |
| "learning_rate": 1.1298405466970387e-05, | |
| "loss": 0.3658, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.38335102796554565, | |
| "step": 125, | |
| "valid_targets_mean": 2847.5, | |
| "valid_targets_min": 570 | |
| }, | |
| { | |
| "epoch": 0.20733652312599682, | |
| "grad_norm": 0.6712559232045431, | |
| "learning_rate": 1.1753986332574032e-05, | |
| "loss": 0.3667, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.37891942262649536, | |
| "step": 130, | |
| "valid_targets_mean": 3401.6, | |
| "valid_targets_min": 1456 | |
| }, | |
| { | |
| "epoch": 0.215311004784689, | |
| "grad_norm": 0.6836972272176005, | |
| "learning_rate": 1.2209567198177677e-05, | |
| "loss": 0.3686, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.41928017139434814, | |
| "step": 135, | |
| "valid_targets_mean": 3157.6, | |
| "valid_targets_min": 471 | |
| }, | |
| { | |
| "epoch": 0.22328548644338117, | |
| "grad_norm": 0.6997393115915425, | |
| "learning_rate": 1.2665148063781323e-05, | |
| "loss": 0.3738, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.34178513288497925, | |
| "step": 140, | |
| "valid_targets_mean": 2723.2, | |
| "valid_targets_min": 1371 | |
| }, | |
| { | |
| "epoch": 0.23125996810207336, | |
| "grad_norm": 0.7360428127205939, | |
| "learning_rate": 1.3120728929384968e-05, | |
| "loss": 0.3395, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.336765855550766, | |
| "step": 145, | |
| "valid_targets_mean": 2483.8, | |
| "valid_targets_min": 1296 | |
| }, | |
| { | |
| "epoch": 0.23923444976076555, | |
| "grad_norm": 0.6300114527878021, | |
| "learning_rate": 1.357630979498861e-05, | |
| "loss": 0.3264, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3560517430305481, | |
| "step": 150, | |
| "valid_targets_mean": 3368.4, | |
| "valid_targets_min": 946 | |
| }, | |
| { | |
| "epoch": 0.24720893141945774, | |
| "grad_norm": 0.7089495360173708, | |
| "learning_rate": 1.4031890660592255e-05, | |
| "loss": 0.3514, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.37352272868156433, | |
| "step": 155, | |
| "valid_targets_mean": 3006.2, | |
| "valid_targets_min": 1037 | |
| }, | |
| { | |
| "epoch": 0.2551834130781499, | |
| "grad_norm": 0.6816824704537204, | |
| "learning_rate": 1.4487471526195902e-05, | |
| "loss": 0.3268, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.31338217854499817, | |
| "step": 160, | |
| "valid_targets_mean": 2765.8, | |
| "valid_targets_min": 1271 | |
| }, | |
| { | |
| "epoch": 0.2631578947368421, | |
| "grad_norm": 0.6672669398637605, | |
| "learning_rate": 1.4943052391799546e-05, | |
| "loss": 0.3408, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3209226131439209, | |
| "step": 165, | |
| "valid_targets_mean": 3222.7, | |
| "valid_targets_min": 1133 | |
| }, | |
| { | |
| "epoch": 0.2711323763955343, | |
| "grad_norm": 0.7103134650450313, | |
| "learning_rate": 1.539863325740319e-05, | |
| "loss": 0.3101, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3130913972854614, | |
| "step": 170, | |
| "valid_targets_mean": 2424.1, | |
| "valid_targets_min": 533 | |
| }, | |
| { | |
| "epoch": 0.27910685805422647, | |
| "grad_norm": 0.8003799128827424, | |
| "learning_rate": 1.5854214123006836e-05, | |
| "loss": 0.293, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27181631326675415, | |
| "step": 175, | |
| "valid_targets_mean": 2400.7, | |
| "valid_targets_min": 1025 | |
| }, | |
| { | |
| "epoch": 0.28708133971291866, | |
| "grad_norm": 0.6232480405624014, | |
| "learning_rate": 1.630979498861048e-05, | |
| "loss": 0.3295, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3603177070617676, | |
| "step": 180, | |
| "valid_targets_mean": 3960.4, | |
| "valid_targets_min": 1658 | |
| }, | |
| { | |
| "epoch": 0.29505582137161085, | |
| "grad_norm": 0.7699746809628365, | |
| "learning_rate": 1.6765375854214125e-05, | |
| "loss": 0.3568, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.31741786003112793, | |
| "step": 185, | |
| "valid_targets_mean": 2706.6, | |
| "valid_targets_min": 1017 | |
| }, | |
| { | |
| "epoch": 0.30303030303030304, | |
| "grad_norm": 0.6458959798755932, | |
| "learning_rate": 1.722095671981777e-05, | |
| "loss": 0.301, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3305377960205078, | |
| "step": 190, | |
| "valid_targets_mean": 3139.9, | |
| "valid_targets_min": 1204 | |
| }, | |
| { | |
| "epoch": 0.31100478468899523, | |
| "grad_norm": 0.716800670588027, | |
| "learning_rate": 1.7676537585421415e-05, | |
| "loss": 0.3094, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.34545591473579407, | |
| "step": 195, | |
| "valid_targets_mean": 2778.2, | |
| "valid_targets_min": 1294 | |
| }, | |
| { | |
| "epoch": 0.3189792663476874, | |
| "grad_norm": 0.7022017213023772, | |
| "learning_rate": 1.813211845102506e-05, | |
| "loss": 0.3211, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.39925310015678406, | |
| "step": 200, | |
| "valid_targets_mean": 3597.8, | |
| "valid_targets_min": 1109 | |
| }, | |
| { | |
| "epoch": 0.3269537480063796, | |
| "grad_norm": 0.6415795025905041, | |
| "learning_rate": 1.8587699316628704e-05, | |
| "loss": 0.3136, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.29438742995262146, | |
| "step": 205, | |
| "valid_targets_mean": 3129.2, | |
| "valid_targets_min": 1143 | |
| }, | |
| { | |
| "epoch": 0.3349282296650718, | |
| "grad_norm": 0.7778916243584333, | |
| "learning_rate": 1.904328018223235e-05, | |
| "loss": 0.2938, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26425981521606445, | |
| "step": 210, | |
| "valid_targets_mean": 2129.7, | |
| "valid_targets_min": 1130 | |
| }, | |
| { | |
| "epoch": 0.34290271132376393, | |
| "grad_norm": 0.6219565329305242, | |
| "learning_rate": 1.9498861047835993e-05, | |
| "loss": 0.3174, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.4421326518058777, | |
| "step": 215, | |
| "valid_targets_mean": 3950.2, | |
| "valid_targets_min": 1277 | |
| }, | |
| { | |
| "epoch": 0.3508771929824561, | |
| "grad_norm": 0.785836372040567, | |
| "learning_rate": 1.9954441913439638e-05, | |
| "loss": 0.29, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.34666532278060913, | |
| "step": 220, | |
| "valid_targets_mean": 2993.6, | |
| "valid_targets_min": 1001 | |
| }, | |
| { | |
| "epoch": 0.3588516746411483, | |
| "grad_norm": 0.746881160297121, | |
| "learning_rate": 2.0410022779043283e-05, | |
| "loss": 0.3313, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.286319375038147, | |
| "step": 225, | |
| "valid_targets_mean": 2257.7, | |
| "valid_targets_min": 1234 | |
| }, | |
| { | |
| "epoch": 0.3668261562998405, | |
| "grad_norm": 0.7507845800604592, | |
| "learning_rate": 2.0865603644646927e-05, | |
| "loss": 0.3111, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.33375227451324463, | |
| "step": 230, | |
| "valid_targets_mean": 2517.9, | |
| "valid_targets_min": 1051 | |
| }, | |
| { | |
| "epoch": 0.3748006379585327, | |
| "grad_norm": 0.6392812481173415, | |
| "learning_rate": 2.1321184510250572e-05, | |
| "loss": 0.298, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.36269882321357727, | |
| "step": 235, | |
| "valid_targets_mean": 3927.3, | |
| "valid_targets_min": 1231 | |
| }, | |
| { | |
| "epoch": 0.3827751196172249, | |
| "grad_norm": 0.7350944642779489, | |
| "learning_rate": 2.1776765375854217e-05, | |
| "loss": 0.2944, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2749019265174866, | |
| "step": 240, | |
| "valid_targets_mean": 2795.9, | |
| "valid_targets_min": 1408 | |
| }, | |
| { | |
| "epoch": 0.39074960127591707, | |
| "grad_norm": 0.6799973116920671, | |
| "learning_rate": 2.223234624145786e-05, | |
| "loss": 0.328, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3284507989883423, | |
| "step": 245, | |
| "valid_targets_mean": 3278.3, | |
| "valid_targets_min": 1179 | |
| }, | |
| { | |
| "epoch": 0.39872408293460926, | |
| "grad_norm": 0.6937881723620776, | |
| "learning_rate": 2.2687927107061506e-05, | |
| "loss": 0.2755, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26359331607818604, | |
| "step": 250, | |
| "valid_targets_mean": 2635.9, | |
| "valid_targets_min": 1254 | |
| }, | |
| { | |
| "epoch": 0.40669856459330145, | |
| "grad_norm": 0.6768083264850395, | |
| "learning_rate": 2.314350797266515e-05, | |
| "loss": 0.3128, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2547405958175659, | |
| "step": 255, | |
| "valid_targets_mean": 2782.8, | |
| "valid_targets_min": 1098 | |
| }, | |
| { | |
| "epoch": 0.41467304625199364, | |
| "grad_norm": 0.6577985129097631, | |
| "learning_rate": 2.3599088838268792e-05, | |
| "loss": 0.273, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27426400780677795, | |
| "step": 260, | |
| "valid_targets_mean": 2750.0, | |
| "valid_targets_min": 1168 | |
| }, | |
| { | |
| "epoch": 0.4226475279106858, | |
| "grad_norm": 0.682947460712769, | |
| "learning_rate": 2.4054669703872436e-05, | |
| "loss": 0.285, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.30281588435173035, | |
| "step": 265, | |
| "valid_targets_mean": 3342.2, | |
| "valid_targets_min": 1294 | |
| }, | |
| { | |
| "epoch": 0.430622009569378, | |
| "grad_norm": 0.746329268810735, | |
| "learning_rate": 2.4510250569476085e-05, | |
| "loss": 0.285, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27677425742149353, | |
| "step": 270, | |
| "valid_targets_mean": 2662.0, | |
| "valid_targets_min": 1108 | |
| }, | |
| { | |
| "epoch": 0.43859649122807015, | |
| "grad_norm": 0.6987593054550297, | |
| "learning_rate": 2.496583143507973e-05, | |
| "loss": 0.3044, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.29885128140449524, | |
| "step": 275, | |
| "valid_targets_mean": 2668.7, | |
| "valid_targets_min": 1410 | |
| }, | |
| { | |
| "epoch": 0.44657097288676234, | |
| "grad_norm": 0.7260544620953092, | |
| "learning_rate": 2.5421412300683374e-05, | |
| "loss": 0.2868, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.29124659299850464, | |
| "step": 280, | |
| "valid_targets_mean": 2561.8, | |
| "valid_targets_min": 1285 | |
| }, | |
| { | |
| "epoch": 0.45454545454545453, | |
| "grad_norm": 0.7281604766084092, | |
| "learning_rate": 2.587699316628702e-05, | |
| "loss": 0.2721, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2808125615119934, | |
| "step": 285, | |
| "valid_targets_mean": 2608.4, | |
| "valid_targets_min": 1218 | |
| }, | |
| { | |
| "epoch": 0.4625199362041467, | |
| "grad_norm": 0.7025214327060316, | |
| "learning_rate": 2.6332574031890663e-05, | |
| "loss": 0.2878, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.31319329142570496, | |
| "step": 290, | |
| "valid_targets_mean": 2582.1, | |
| "valid_targets_min": 1187 | |
| }, | |
| { | |
| "epoch": 0.4704944178628389, | |
| "grad_norm": 0.7856192992209702, | |
| "learning_rate": 2.6788154897494308e-05, | |
| "loss": 0.2823, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27091526985168457, | |
| "step": 295, | |
| "valid_targets_mean": 2811.6, | |
| "valid_targets_min": 624 | |
| }, | |
| { | |
| "epoch": 0.4784688995215311, | |
| "grad_norm": 0.7108315007493442, | |
| "learning_rate": 2.7243735763097953e-05, | |
| "loss": 0.3031, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3215833902359009, | |
| "step": 300, | |
| "valid_targets_mean": 2883.8, | |
| "valid_targets_min": 1066 | |
| }, | |
| { | |
| "epoch": 0.4864433811802233, | |
| "grad_norm": 0.7748559874677846, | |
| "learning_rate": 2.7699316628701597e-05, | |
| "loss": 0.2579, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22398757934570312, | |
| "step": 305, | |
| "valid_targets_mean": 2005.8, | |
| "valid_targets_min": 1123 | |
| }, | |
| { | |
| "epoch": 0.4944178628389155, | |
| "grad_norm": 0.7462878729201242, | |
| "learning_rate": 2.815489749430524e-05, | |
| "loss": 0.3319, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.33022865653038025, | |
| "step": 310, | |
| "valid_targets_mean": 2705.2, | |
| "valid_targets_min": 248 | |
| }, | |
| { | |
| "epoch": 0.5023923444976076, | |
| "grad_norm": 0.7802542037539941, | |
| "learning_rate": 2.8610478359908883e-05, | |
| "loss": 0.2839, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23213829100131989, | |
| "step": 315, | |
| "valid_targets_mean": 2407.5, | |
| "valid_targets_min": 1251 | |
| }, | |
| { | |
| "epoch": 0.5103668261562998, | |
| "grad_norm": 0.7649747995292412, | |
| "learning_rate": 2.906605922551253e-05, | |
| "loss": 0.2961, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3233932852745056, | |
| "step": 320, | |
| "valid_targets_mean": 2915.9, | |
| "valid_targets_min": 1207 | |
| }, | |
| { | |
| "epoch": 0.518341307814992, | |
| "grad_norm": 0.7978888829804283, | |
| "learning_rate": 2.9521640091116176e-05, | |
| "loss": 0.276, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22950753569602966, | |
| "step": 325, | |
| "valid_targets_mean": 2392.1, | |
| "valid_targets_min": 987 | |
| }, | |
| { | |
| "epoch": 0.5263157894736842, | |
| "grad_norm": 0.7437520517670876, | |
| "learning_rate": 2.997722095671982e-05, | |
| "loss": 0.2964, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2997945249080658, | |
| "step": 330, | |
| "valid_targets_mean": 3249.1, | |
| "valid_targets_min": 1234 | |
| }, | |
| { | |
| "epoch": 0.5342902711323764, | |
| "grad_norm": 0.6657204266063321, | |
| "learning_rate": 3.0432801822323465e-05, | |
| "loss": 0.2606, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.29147499799728394, | |
| "step": 335, | |
| "valid_targets_mean": 3084.5, | |
| "valid_targets_min": 1182 | |
| }, | |
| { | |
| "epoch": 0.5422647527910686, | |
| "grad_norm": 0.605457899238917, | |
| "learning_rate": 3.088838268792711e-05, | |
| "loss": 0.2629, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23867332935333252, | |
| "step": 340, | |
| "valid_targets_mean": 3086.1, | |
| "valid_targets_min": 1255 | |
| }, | |
| { | |
| "epoch": 0.5502392344497608, | |
| "grad_norm": 0.677848238624908, | |
| "learning_rate": 3.1343963553530755e-05, | |
| "loss": 0.2957, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3328245282173157, | |
| "step": 345, | |
| "valid_targets_mean": 2985.1, | |
| "valid_targets_min": 1422 | |
| }, | |
| { | |
| "epoch": 0.5582137161084529, | |
| "grad_norm": 0.6938963916271617, | |
| "learning_rate": 3.17995444191344e-05, | |
| "loss": 0.2893, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2992449998855591, | |
| "step": 350, | |
| "valid_targets_mean": 3391.4, | |
| "valid_targets_min": 899 | |
| }, | |
| { | |
| "epoch": 0.5661881977671451, | |
| "grad_norm": 0.6671036083494569, | |
| "learning_rate": 3.2255125284738044e-05, | |
| "loss": 0.2639, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.261359304189682, | |
| "step": 355, | |
| "valid_targets_mean": 2705.2, | |
| "valid_targets_min": 1199 | |
| }, | |
| { | |
| "epoch": 0.5741626794258373, | |
| "grad_norm": 0.641793075175834, | |
| "learning_rate": 3.271070615034169e-05, | |
| "loss": 0.2563, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2564861476421356, | |
| "step": 360, | |
| "valid_targets_mean": 3231.8, | |
| "valid_targets_min": 1258 | |
| }, | |
| { | |
| "epoch": 0.5821371610845295, | |
| "grad_norm": 0.6260519357718421, | |
| "learning_rate": 3.316628701594533e-05, | |
| "loss": 0.2795, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.29788124561309814, | |
| "step": 365, | |
| "valid_targets_mean": 3343.8, | |
| "valid_targets_min": 998 | |
| }, | |
| { | |
| "epoch": 0.5901116427432217, | |
| "grad_norm": 0.7016875050444317, | |
| "learning_rate": 3.362186788154898e-05, | |
| "loss": 0.2625, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23810943961143494, | |
| "step": 370, | |
| "valid_targets_mean": 2345.8, | |
| "valid_targets_min": 1102 | |
| }, | |
| { | |
| "epoch": 0.5980861244019139, | |
| "grad_norm": 0.6940263992622621, | |
| "learning_rate": 3.407744874715262e-05, | |
| "loss": 0.2589, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.239394873380661, | |
| "step": 375, | |
| "valid_targets_mean": 2608.0, | |
| "valid_targets_min": 1286 | |
| }, | |
| { | |
| "epoch": 0.6060606060606061, | |
| "grad_norm": 0.7346814637270078, | |
| "learning_rate": 3.453302961275627e-05, | |
| "loss": 0.2878, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3815971612930298, | |
| "step": 380, | |
| "valid_targets_mean": 3218.8, | |
| "valid_targets_min": 1093 | |
| }, | |
| { | |
| "epoch": 0.6140350877192983, | |
| "grad_norm": 0.6058315721294318, | |
| "learning_rate": 3.498861047835991e-05, | |
| "loss": 0.2755, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2890080213546753, | |
| "step": 385, | |
| "valid_targets_mean": 3851.6, | |
| "valid_targets_min": 1095 | |
| }, | |
| { | |
| "epoch": 0.6220095693779905, | |
| "grad_norm": 0.7023734773136625, | |
| "learning_rate": 3.5444191343963557e-05, | |
| "loss": 0.2706, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.30273354053497314, | |
| "step": 390, | |
| "valid_targets_mean": 2969.1, | |
| "valid_targets_min": 1399 | |
| }, | |
| { | |
| "epoch": 0.6299840510366826, | |
| "grad_norm": 1.4920526075234164, | |
| "learning_rate": 3.58997722095672e-05, | |
| "loss": 0.2762, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22971850633621216, | |
| "step": 395, | |
| "valid_targets_mean": 2233.4, | |
| "valid_targets_min": 1057 | |
| }, | |
| { | |
| "epoch": 0.6379585326953748, | |
| "grad_norm": 0.6703882371109363, | |
| "learning_rate": 3.6355353075170846e-05, | |
| "loss": 0.2807, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.30465108156204224, | |
| "step": 400, | |
| "valid_targets_mean": 2949.6, | |
| "valid_targets_min": 1207 | |
| }, | |
| { | |
| "epoch": 0.645933014354067, | |
| "grad_norm": 0.6301067583808945, | |
| "learning_rate": 3.681093394077449e-05, | |
| "loss": 0.2557, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2273084968328476, | |
| "step": 405, | |
| "valid_targets_mean": 2778.0, | |
| "valid_targets_min": 1153 | |
| }, | |
| { | |
| "epoch": 0.6539074960127592, | |
| "grad_norm": 0.7296897171202118, | |
| "learning_rate": 3.7266514806378135e-05, | |
| "loss": 0.235, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27479881048202515, | |
| "step": 410, | |
| "valid_targets_mean": 2959.2, | |
| "valid_targets_min": 1149 | |
| }, | |
| { | |
| "epoch": 0.6618819776714514, | |
| "grad_norm": 0.6281384576410912, | |
| "learning_rate": 3.772209567198178e-05, | |
| "loss": 0.2755, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25720399618148804, | |
| "step": 415, | |
| "valid_targets_mean": 3779.6, | |
| "valid_targets_min": 1287 | |
| }, | |
| { | |
| "epoch": 0.6698564593301436, | |
| "grad_norm": 0.5853097243857709, | |
| "learning_rate": 3.8177676537585425e-05, | |
| "loss": 0.258, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3213130533695221, | |
| "step": 420, | |
| "valid_targets_mean": 4250.0, | |
| "valid_targets_min": 1513 | |
| }, | |
| { | |
| "epoch": 0.6778309409888357, | |
| "grad_norm": 0.7301060156034734, | |
| "learning_rate": 3.863325740318907e-05, | |
| "loss": 0.251, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26980870962142944, | |
| "step": 425, | |
| "valid_targets_mean": 2502.8, | |
| "valid_targets_min": 1158 | |
| }, | |
| { | |
| "epoch": 0.6858054226475279, | |
| "grad_norm": 0.5832615343218808, | |
| "learning_rate": 3.9088838268792714e-05, | |
| "loss": 0.2594, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2549774646759033, | |
| "step": 430, | |
| "valid_targets_mean": 3444.6, | |
| "valid_targets_min": 998 | |
| }, | |
| { | |
| "epoch": 0.69377990430622, | |
| "grad_norm": 0.749747412812053, | |
| "learning_rate": 3.954441913439636e-05, | |
| "loss": 0.2566, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2809187173843384, | |
| "step": 435, | |
| "valid_targets_mean": 2738.5, | |
| "valid_targets_min": 1269 | |
| }, | |
| { | |
| "epoch": 0.7017543859649122, | |
| "grad_norm": 0.6682751126395767, | |
| "learning_rate": 4e-05, | |
| "loss": 0.247, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24001020193099976, | |
| "step": 440, | |
| "valid_targets_mean": 2767.5, | |
| "valid_targets_min": 1374 | |
| }, | |
| { | |
| "epoch": 0.7097288676236044, | |
| "grad_norm": 0.647984542933624, | |
| "learning_rate": 3.9999841858814384e-05, | |
| "loss": 0.3028, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.29405590891838074, | |
| "step": 445, | |
| "valid_targets_mean": 3640.1, | |
| "valid_targets_min": 1438 | |
| }, | |
| { | |
| "epoch": 0.7177033492822966, | |
| "grad_norm": 0.6803155407263939, | |
| "learning_rate": 3.999936743775839e-05, | |
| "loss": 0.283, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2493191361427307, | |
| "step": 450, | |
| "valid_targets_mean": 2872.7, | |
| "valid_targets_min": 1156 | |
| }, | |
| { | |
| "epoch": 0.7256778309409888, | |
| "grad_norm": 0.5860939981094048, | |
| "learning_rate": 3.9998576744334574e-05, | |
| "loss": 0.2728, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.31388020515441895, | |
| "step": 455, | |
| "valid_targets_mean": 4003.9, | |
| "valid_targets_min": 1248 | |
| }, | |
| { | |
| "epoch": 0.733652312599681, | |
| "grad_norm": 0.7082359097380168, | |
| "learning_rate": 3.999746979104705e-05, | |
| "loss": 0.2537, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.265491247177124, | |
| "step": 460, | |
| "valid_targets_mean": 2565.0, | |
| "valid_targets_min": 1501 | |
| }, | |
| { | |
| "epoch": 0.7416267942583732, | |
| "grad_norm": 0.7642274897132844, | |
| "learning_rate": 3.999604659540131e-05, | |
| "loss": 0.2751, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.284295916557312, | |
| "step": 465, | |
| "valid_targets_mean": 3430.8, | |
| "valid_targets_min": 985 | |
| }, | |
| { | |
| "epoch": 0.7496012759170654, | |
| "grad_norm": 0.5893515909471992, | |
| "learning_rate": 3.999430717990395e-05, | |
| "loss": 0.2728, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2829861640930176, | |
| "step": 470, | |
| "valid_targets_mean": 3407.2, | |
| "valid_targets_min": 1353 | |
| }, | |
| { | |
| "epoch": 0.7575757575757576, | |
| "grad_norm": 0.8136772121235736, | |
| "learning_rate": 3.999225157206228e-05, | |
| "loss": 0.2526, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23452086746692657, | |
| "step": 475, | |
| "valid_targets_mean": 1940.5, | |
| "valid_targets_min": 1060 | |
| }, | |
| { | |
| "epoch": 0.7655502392344498, | |
| "grad_norm": 0.7188632991013135, | |
| "learning_rate": 3.998987980438393e-05, | |
| "loss": 0.251, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19779305160045624, | |
| "step": 480, | |
| "valid_targets_mean": 2111.4, | |
| "valid_targets_min": 1441 | |
| }, | |
| { | |
| "epoch": 0.773524720893142, | |
| "grad_norm": 0.5775811168461635, | |
| "learning_rate": 3.9987191914376306e-05, | |
| "loss": 0.262, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2576785683631897, | |
| "step": 485, | |
| "valid_targets_mean": 3279.2, | |
| "valid_targets_min": 1051 | |
| }, | |
| { | |
| "epoch": 0.7814992025518341, | |
| "grad_norm": 0.7007873699634931, | |
| "learning_rate": 3.998418794454604e-05, | |
| "loss": 0.2439, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2803601920604706, | |
| "step": 490, | |
| "valid_targets_mean": 3080.1, | |
| "valid_targets_min": 1657 | |
| }, | |
| { | |
| "epoch": 0.7894736842105263, | |
| "grad_norm": 0.6007025928633855, | |
| "learning_rate": 3.998086794239825e-05, | |
| "loss": 0.2829, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.32280203700065613, | |
| "step": 495, | |
| "valid_targets_mean": 4255.9, | |
| "valid_targets_min": 1680 | |
| }, | |
| { | |
| "epoch": 0.7974481658692185, | |
| "grad_norm": 0.6880645344758224, | |
| "learning_rate": 3.997723196043585e-05, | |
| "loss": 0.2605, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24724368751049042, | |
| "step": 500, | |
| "valid_targets_mean": 2747.2, | |
| "valid_targets_min": 1084 | |
| }, | |
| { | |
| "epoch": 0.8054226475279107, | |
| "grad_norm": 0.6773462313419402, | |
| "learning_rate": 3.9973280056158695e-05, | |
| "loss": 0.2426, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2253529280424118, | |
| "step": 505, | |
| "valid_targets_mean": 2516.2, | |
| "valid_targets_min": 1166 | |
| }, | |
| { | |
| "epoch": 0.8133971291866029, | |
| "grad_norm": 0.7173069626107433, | |
| "learning_rate": 3.9969012292062655e-05, | |
| "loss": 0.2671, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27754151821136475, | |
| "step": 510, | |
| "valid_targets_mean": 2236.1, | |
| "valid_targets_min": 269 | |
| }, | |
| { | |
| "epoch": 0.8213716108452951, | |
| "grad_norm": 0.6542849868717581, | |
| "learning_rate": 3.996442873563866e-05, | |
| "loss": 0.2611, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24865597486495972, | |
| "step": 515, | |
| "valid_targets_mean": 2807.8, | |
| "valid_targets_min": 1347 | |
| }, | |
| { | |
| "epoch": 0.8293460925039873, | |
| "grad_norm": 0.6023761211547769, | |
| "learning_rate": 3.9959529459371624e-05, | |
| "loss": 0.2549, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24921664595603943, | |
| "step": 520, | |
| "valid_targets_mean": 2813.4, | |
| "valid_targets_min": 1219 | |
| }, | |
| { | |
| "epoch": 0.8373205741626795, | |
| "grad_norm": 0.688792403344696, | |
| "learning_rate": 3.9954314540739284e-05, | |
| "loss": 0.2577, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2918543219566345, | |
| "step": 525, | |
| "valid_targets_mean": 2673.4, | |
| "valid_targets_min": 920 | |
| }, | |
| { | |
| "epoch": 0.8452950558213717, | |
| "grad_norm": 0.5571463923330057, | |
| "learning_rate": 3.994878406221097e-05, | |
| "loss": 0.2576, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.31798291206359863, | |
| "step": 530, | |
| "valid_targets_mean": 3787.4, | |
| "valid_targets_min": 965 | |
| }, | |
| { | |
| "epoch": 0.8532695374800638, | |
| "grad_norm": 0.5343440337170275, | |
| "learning_rate": 3.994293811124632e-05, | |
| "loss": 0.2699, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3140130043029785, | |
| "step": 535, | |
| "valid_targets_mean": 4513.2, | |
| "valid_targets_min": 1192 | |
| }, | |
| { | |
| "epoch": 0.861244019138756, | |
| "grad_norm": 0.6667148523666464, | |
| "learning_rate": 3.993677678029392e-05, | |
| "loss": 0.2883, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2631716728210449, | |
| "step": 540, | |
| "valid_targets_mean": 2576.0, | |
| "valid_targets_min": 1248 | |
| }, | |
| { | |
| "epoch": 0.8692185007974481, | |
| "grad_norm": 0.621797201029701, | |
| "learning_rate": 3.9930300166789765e-05, | |
| "loss": 0.2529, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23845577239990234, | |
| "step": 545, | |
| "valid_targets_mean": 2497.0, | |
| "valid_targets_min": 1015 | |
| }, | |
| { | |
| "epoch": 0.8771929824561403, | |
| "grad_norm": 0.5849281982647846, | |
| "learning_rate": 3.992350837315581e-05, | |
| "loss": 0.2403, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24086298048496246, | |
| "step": 550, | |
| "valid_targets_mean": 2936.5, | |
| "valid_targets_min": 1088 | |
| }, | |
| { | |
| "epoch": 0.8851674641148325, | |
| "grad_norm": 0.6100603750245858, | |
| "learning_rate": 3.991640150679826e-05, | |
| "loss": 0.2442, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2142607867717743, | |
| "step": 555, | |
| "valid_targets_mean": 2693.3, | |
| "valid_targets_min": 996 | |
| }, | |
| { | |
| "epoch": 0.8931419457735247, | |
| "grad_norm": 0.679595548377651, | |
| "learning_rate": 3.990897968010596e-05, | |
| "loss": 0.2551, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23986729979515076, | |
| "step": 560, | |
| "valid_targets_mean": 2657.9, | |
| "valid_targets_min": 1230 | |
| }, | |
| { | |
| "epoch": 0.9011164274322169, | |
| "grad_norm": 0.5714798003175021, | |
| "learning_rate": 3.990124301044855e-05, | |
| "loss": 0.2359, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22235023975372314, | |
| "step": 565, | |
| "valid_targets_mean": 3395.3, | |
| "valid_targets_min": 1042 | |
| }, | |
| { | |
| "epoch": 0.9090909090909091, | |
| "grad_norm": 0.6184667605667801, | |
| "learning_rate": 3.989319162017465e-05, | |
| "loss": 0.2446, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22823108732700348, | |
| "step": 570, | |
| "valid_targets_mean": 2967.4, | |
| "valid_targets_min": 1564 | |
| }, | |
| { | |
| "epoch": 0.9170653907496013, | |
| "grad_norm": 0.7112571643134058, | |
| "learning_rate": 3.988482563660989e-05, | |
| "loss": 0.2718, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.30855730175971985, | |
| "step": 575, | |
| "valid_targets_mean": 2733.8, | |
| "valid_targets_min": 753 | |
| }, | |
| { | |
| "epoch": 0.9250398724082934, | |
| "grad_norm": 0.5409940557899795, | |
| "learning_rate": 3.987614519205493e-05, | |
| "loss": 0.2587, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2341741919517517, | |
| "step": 580, | |
| "valid_targets_mean": 3420.3, | |
| "valid_targets_min": 1238 | |
| }, | |
| { | |
| "epoch": 0.9330143540669856, | |
| "grad_norm": 0.6283821372972509, | |
| "learning_rate": 3.986715042378334e-05, | |
| "loss": 0.2549, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26625967025756836, | |
| "step": 585, | |
| "valid_targets_mean": 2681.2, | |
| "valid_targets_min": 1477 | |
| }, | |
| { | |
| "epoch": 0.9409888357256778, | |
| "grad_norm": 0.5958617621432479, | |
| "learning_rate": 3.985784147403947e-05, | |
| "loss": 0.2227, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2559368312358856, | |
| "step": 590, | |
| "valid_targets_mean": 2903.1, | |
| "valid_targets_min": 1321 | |
| }, | |
| { | |
| "epoch": 0.94896331738437, | |
| "grad_norm": 0.5949783472882877, | |
| "learning_rate": 3.9848218490036144e-05, | |
| "loss": 0.2478, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2529694437980652, | |
| "step": 595, | |
| "valid_targets_mean": 3612.6, | |
| "valid_targets_min": 1114 | |
| }, | |
| { | |
| "epoch": 0.9569377990430622, | |
| "grad_norm": 0.5701025918722548, | |
| "learning_rate": 3.983828162395238e-05, | |
| "loss": 0.2505, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24346418678760529, | |
| "step": 600, | |
| "valid_targets_mean": 3209.6, | |
| "valid_targets_min": 1324 | |
| }, | |
| { | |
| "epoch": 0.9649122807017544, | |
| "grad_norm": 1.3475027154750536, | |
| "learning_rate": 3.9828031032930944e-05, | |
| "loss": 0.2693, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23624694347381592, | |
| "step": 605, | |
| "valid_targets_mean": 2403.8, | |
| "valid_targets_min": 1275 | |
| }, | |
| { | |
| "epoch": 0.9728867623604466, | |
| "grad_norm": 0.6783942654321022, | |
| "learning_rate": 3.98174668790759e-05, | |
| "loss": 0.2531, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28323280811309814, | |
| "step": 610, | |
| "valid_targets_mean": 3008.2, | |
| "valid_targets_min": 1333 | |
| }, | |
| { | |
| "epoch": 0.9808612440191388, | |
| "grad_norm": 0.7285748900158596, | |
| "learning_rate": 3.9806589329450045e-05, | |
| "loss": 0.2336, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23608598113059998, | |
| "step": 615, | |
| "valid_targets_mean": 1979.1, | |
| "valid_targets_min": 1174 | |
| }, | |
| { | |
| "epoch": 0.988835725677831, | |
| "grad_norm": 0.5779028885760018, | |
| "learning_rate": 3.979539855607222e-05, | |
| "loss": 0.2514, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22704467177391052, | |
| "step": 620, | |
| "valid_targets_mean": 2778.1, | |
| "valid_targets_min": 719 | |
| }, | |
| { | |
| "epoch": 0.9968102073365231, | |
| "grad_norm": 0.53910671747115, | |
| "learning_rate": 3.9783894735914646e-05, | |
| "loss": 0.2609, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24189332127571106, | |
| "step": 625, | |
| "valid_targets_mean": 3171.7, | |
| "valid_targets_min": 684 | |
| }, | |
| { | |
| "epoch": 1.0047846889952152, | |
| "grad_norm": 0.5590992427797843, | |
| "learning_rate": 3.9772078050900105e-05, | |
| "loss": 0.2519, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.267245352268219, | |
| "step": 630, | |
| "valid_targets_mean": 3224.1, | |
| "valid_targets_min": 1330 | |
| }, | |
| { | |
| "epoch": 1.0127591706539074, | |
| "grad_norm": 0.5412219374722034, | |
| "learning_rate": 3.9759948687899055e-05, | |
| "loss": 0.2367, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2462998479604721, | |
| "step": 635, | |
| "valid_targets_mean": 3618.4, | |
| "valid_targets_min": 1397 | |
| }, | |
| { | |
| "epoch": 1.0207336523125996, | |
| "grad_norm": 0.6043980742100203, | |
| "learning_rate": 3.974750683872667e-05, | |
| "loss": 0.2515, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24493081867694855, | |
| "step": 640, | |
| "valid_targets_mean": 2639.1, | |
| "valid_targets_min": 1182 | |
| }, | |
| { | |
| "epoch": 1.0287081339712918, | |
| "grad_norm": 0.7096444272979168, | |
| "learning_rate": 3.973475270013984e-05, | |
| "loss": 0.2498, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.33669358491897583, | |
| "step": 645, | |
| "valid_targets_mean": 2667.2, | |
| "valid_targets_min": 1023 | |
| }, | |
| { | |
| "epoch": 1.036682615629984, | |
| "grad_norm": 0.5889749694176992, | |
| "learning_rate": 3.972168647383402e-05, | |
| "loss": 0.2135, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20457293093204498, | |
| "step": 650, | |
| "valid_targets_mean": 2950.1, | |
| "valid_targets_min": 803 | |
| }, | |
| { | |
| "epoch": 1.0446570972886762, | |
| "grad_norm": 0.545750879813732, | |
| "learning_rate": 3.970830836644006e-05, | |
| "loss": 0.2317, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23334968090057373, | |
| "step": 655, | |
| "valid_targets_mean": 3426.9, | |
| "valid_targets_min": 1214 | |
| }, | |
| { | |
| "epoch": 1.0526315789473684, | |
| "grad_norm": 0.6522530149915291, | |
| "learning_rate": 3.9694618589520945e-05, | |
| "loss": 0.202, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21202175319194794, | |
| "step": 660, | |
| "valid_targets_mean": 2608.2, | |
| "valid_targets_min": 1123 | |
| }, | |
| { | |
| "epoch": 1.0606060606060606, | |
| "grad_norm": 0.6005460698394031, | |
| "learning_rate": 3.9680617359568414e-05, | |
| "loss": 0.2469, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2296278476715088, | |
| "step": 665, | |
| "valid_targets_mean": 2712.1, | |
| "valid_targets_min": 930 | |
| }, | |
| { | |
| "epoch": 1.0685805422647527, | |
| "grad_norm": 0.6017327323981438, | |
| "learning_rate": 3.966630489799959e-05, | |
| "loss": 0.2103, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21935969591140747, | |
| "step": 670, | |
| "valid_targets_mean": 2787.2, | |
| "valid_targets_min": 1514 | |
| }, | |
| { | |
| "epoch": 1.076555023923445, | |
| "grad_norm": 0.6150691268319541, | |
| "learning_rate": 3.9651681431153445e-05, | |
| "loss": 0.236, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22077104449272156, | |
| "step": 675, | |
| "valid_targets_mean": 2930.1, | |
| "valid_targets_min": 1093 | |
| }, | |
| { | |
| "epoch": 1.0845295055821371, | |
| "grad_norm": 0.6200136891990149, | |
| "learning_rate": 3.96367471902872e-05, | |
| "loss": 0.2118, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19286946952342987, | |
| "step": 680, | |
| "valid_targets_mean": 2750.2, | |
| "valid_targets_min": 1313 | |
| }, | |
| { | |
| "epoch": 1.0925039872408293, | |
| "grad_norm": 0.5655815816732179, | |
| "learning_rate": 3.9621502411572705e-05, | |
| "loss": 0.2595, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24388240277767181, | |
| "step": 685, | |
| "valid_targets_mean": 3086.5, | |
| "valid_targets_min": 1210 | |
| }, | |
| { | |
| "epoch": 1.1004784688995215, | |
| "grad_norm": 0.5885904619456156, | |
| "learning_rate": 3.960594733609273e-05, | |
| "loss": 0.2307, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2468026876449585, | |
| "step": 690, | |
| "valid_targets_mean": 3201.1, | |
| "valid_targets_min": 948 | |
| }, | |
| { | |
| "epoch": 1.1084529505582137, | |
| "grad_norm": 0.5232919415905639, | |
| "learning_rate": 3.9590082209837054e-05, | |
| "loss": 0.2231, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22746863961219788, | |
| "step": 695, | |
| "valid_targets_mean": 3608.3, | |
| "valid_targets_min": 1395 | |
| }, | |
| { | |
| "epoch": 1.1164274322169059, | |
| "grad_norm": 0.5795983343051004, | |
| "learning_rate": 3.957390728369867e-05, | |
| "loss": 0.2181, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21875368058681488, | |
| "step": 700, | |
| "valid_targets_mean": 2985.8, | |
| "valid_targets_min": 1127 | |
| }, | |
| { | |
| "epoch": 1.124401913875598, | |
| "grad_norm": 0.6474674101877196, | |
| "learning_rate": 3.955742281346979e-05, | |
| "loss": 0.2168, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24110597372055054, | |
| "step": 705, | |
| "valid_targets_mean": 2594.4, | |
| "valid_targets_min": 927 | |
| }, | |
| { | |
| "epoch": 1.1323763955342903, | |
| "grad_norm": 0.6120178805218351, | |
| "learning_rate": 3.9540629059837767e-05, | |
| "loss": 0.2199, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2766246795654297, | |
| "step": 710, | |
| "valid_targets_mean": 2957.4, | |
| "valid_targets_min": 1321 | |
| }, | |
| { | |
| "epoch": 1.1403508771929824, | |
| "grad_norm": 0.6448170583460983, | |
| "learning_rate": 3.952352628838102e-05, | |
| "loss": 0.2222, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19483381509780884, | |
| "step": 715, | |
| "valid_targets_mean": 2121.1, | |
| "valid_targets_min": 624 | |
| }, | |
| { | |
| "epoch": 1.1483253588516746, | |
| "grad_norm": 0.6548168327201195, | |
| "learning_rate": 3.95061147695648e-05, | |
| "loss": 0.2324, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21812087297439575, | |
| "step": 720, | |
| "valid_targets_mean": 2455.1, | |
| "valid_targets_min": 1072 | |
| }, | |
| { | |
| "epoch": 1.1562998405103668, | |
| "grad_norm": 0.6190274225455182, | |
| "learning_rate": 3.9488394778736935e-05, | |
| "loss": 0.2746, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22314514219760895, | |
| "step": 725, | |
| "valid_targets_mean": 2731.3, | |
| "valid_targets_min": 1342 | |
| }, | |
| { | |
| "epoch": 1.164274322169059, | |
| "grad_norm": 0.6010823414002561, | |
| "learning_rate": 3.947036659612345e-05, | |
| "loss": 0.2324, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2784811854362488, | |
| "step": 730, | |
| "valid_targets_mean": 3532.8, | |
| "valid_targets_min": 1122 | |
| }, | |
| { | |
| "epoch": 1.1722488038277512, | |
| "grad_norm": 0.6149720156451945, | |
| "learning_rate": 3.945203050682418e-05, | |
| "loss": 0.2362, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2704809308052063, | |
| "step": 735, | |
| "valid_targets_mean": 3149.3, | |
| "valid_targets_min": 984 | |
| }, | |
| { | |
| "epoch": 1.1802232854864434, | |
| "grad_norm": 0.510802694097458, | |
| "learning_rate": 3.94333868008082e-05, | |
| "loss": 0.2161, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2353009134531021, | |
| "step": 740, | |
| "valid_targets_mean": 3687.4, | |
| "valid_targets_min": 1494 | |
| }, | |
| { | |
| "epoch": 1.1881977671451356, | |
| "grad_norm": 0.6004191651529056, | |
| "learning_rate": 3.94144357729093e-05, | |
| "loss": 0.2427, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.27018919587135315, | |
| "step": 745, | |
| "valid_targets_mean": 3065.2, | |
| "valid_targets_min": 1087 | |
| }, | |
| { | |
| "epoch": 1.1961722488038278, | |
| "grad_norm": 0.5173197330016904, | |
| "learning_rate": 3.939517772282127e-05, | |
| "loss": 0.2183, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19749709963798523, | |
| "step": 750, | |
| "valid_targets_mean": 3088.0, | |
| "valid_targets_min": 1262 | |
| }, | |
| { | |
| "epoch": 1.20414673046252, | |
| "grad_norm": 0.5989450936696408, | |
| "learning_rate": 3.93756129550932e-05, | |
| "loss": 0.2236, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22018079459667206, | |
| "step": 755, | |
| "valid_targets_mean": 2998.4, | |
| "valid_targets_min": 1572 | |
| }, | |
| { | |
| "epoch": 1.2121212121212122, | |
| "grad_norm": 0.570268839573639, | |
| "learning_rate": 3.935574177912465e-05, | |
| "loss": 0.2564, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2346849888563156, | |
| "step": 760, | |
| "valid_targets_mean": 3287.2, | |
| "valid_targets_min": 1074 | |
| }, | |
| { | |
| "epoch": 1.2200956937799043, | |
| "grad_norm": 0.601105124747731, | |
| "learning_rate": 3.9335564509160746e-05, | |
| "loss": 0.2372, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2528783977031708, | |
| "step": 765, | |
| "valid_targets_mean": 3077.8, | |
| "valid_targets_min": 991 | |
| }, | |
| { | |
| "epoch": 1.2280701754385965, | |
| "grad_norm": 0.6142966339382259, | |
| "learning_rate": 3.931508146428724e-05, | |
| "loss": 0.223, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22747379541397095, | |
| "step": 770, | |
| "valid_targets_mean": 2825.7, | |
| "valid_targets_min": 1102 | |
| }, | |
| { | |
| "epoch": 1.2360446570972887, | |
| "grad_norm": 0.5087997480632052, | |
| "learning_rate": 3.929429296842542e-05, | |
| "loss": 0.2588, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23831650614738464, | |
| "step": 775, | |
| "valid_targets_mean": 3389.8, | |
| "valid_targets_min": 876 | |
| }, | |
| { | |
| "epoch": 1.244019138755981, | |
| "grad_norm": 0.5239878661727811, | |
| "learning_rate": 3.927319935032703e-05, | |
| "loss": 0.2142, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2023959755897522, | |
| "step": 780, | |
| "valid_targets_mean": 3227.9, | |
| "valid_targets_min": 1260 | |
| }, | |
| { | |
| "epoch": 1.251993620414673, | |
| "grad_norm": 0.6968846240351381, | |
| "learning_rate": 3.925180094356905e-05, | |
| "loss": 0.2242, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20895949006080627, | |
| "step": 785, | |
| "valid_targets_mean": 2142.9, | |
| "valid_targets_min": 1298 | |
| }, | |
| { | |
| "epoch": 1.2599681020733653, | |
| "grad_norm": 0.6041827706842907, | |
| "learning_rate": 3.9230098086548414e-05, | |
| "loss": 0.2136, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19941559433937073, | |
| "step": 790, | |
| "valid_targets_mean": 2695.4, | |
| "valid_targets_min": 1440 | |
| }, | |
| { | |
| "epoch": 1.2679425837320575, | |
| "grad_norm": 0.5775916446397974, | |
| "learning_rate": 3.920809112247668e-05, | |
| "loss": 0.216, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21255138516426086, | |
| "step": 795, | |
| "valid_targets_mean": 2856.7, | |
| "valid_targets_min": 1261 | |
| }, | |
| { | |
| "epoch": 1.2759170653907497, | |
| "grad_norm": 0.5543005602140678, | |
| "learning_rate": 3.918578039937459e-05, | |
| "loss": 0.2442, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2445598691701889, | |
| "step": 800, | |
| "valid_targets_mean": 3451.8, | |
| "valid_targets_min": 1607 | |
| }, | |
| { | |
| "epoch": 1.2838915470494419, | |
| "grad_norm": 0.560785534024231, | |
| "learning_rate": 3.916316627006656e-05, | |
| "loss": 0.2197, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19521670043468475, | |
| "step": 805, | |
| "valid_targets_mean": 3096.6, | |
| "valid_targets_min": 1135 | |
| }, | |
| { | |
| "epoch": 1.291866028708134, | |
| "grad_norm": 0.5767637988086921, | |
| "learning_rate": 3.914024909217511e-05, | |
| "loss": 0.2598, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2539677619934082, | |
| "step": 810, | |
| "valid_targets_mean": 3374.2, | |
| "valid_targets_min": 1374 | |
| }, | |
| { | |
| "epoch": 1.2998405103668262, | |
| "grad_norm": 0.6209716508260346, | |
| "learning_rate": 3.911702922811522e-05, | |
| "loss": 0.2388, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19697776436805725, | |
| "step": 815, | |
| "valid_targets_mean": 2215.6, | |
| "valid_targets_min": 601 | |
| }, | |
| { | |
| "epoch": 1.3078149920255182, | |
| "grad_norm": 0.6291469867776873, | |
| "learning_rate": 3.909350704508856e-05, | |
| "loss": 0.2348, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23938624560832977, | |
| "step": 820, | |
| "valid_targets_mean": 2760.1, | |
| "valid_targets_min": 1112 | |
| }, | |
| { | |
| "epoch": 1.3157894736842106, | |
| "grad_norm": 0.5575921172983946, | |
| "learning_rate": 3.906968291507773e-05, | |
| "loss": 0.2158, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2400011122226715, | |
| "step": 825, | |
| "valid_targets_mean": 3671.2, | |
| "valid_targets_min": 1357 | |
| }, | |
| { | |
| "epoch": 1.3237639553429026, | |
| "grad_norm": 0.6345082402652992, | |
| "learning_rate": 3.904555721484034e-05, | |
| "loss": 0.2426, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24229951202869415, | |
| "step": 830, | |
| "valid_targets_mean": 2401.9, | |
| "valid_targets_min": 1059 | |
| }, | |
| { | |
| "epoch": 1.331738437001595, | |
| "grad_norm": 0.6195684041173591, | |
| "learning_rate": 3.9021130325903076e-05, | |
| "loss": 0.2487, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3127402663230896, | |
| "step": 835, | |
| "valid_targets_mean": 3132.9, | |
| "valid_targets_min": 1253 | |
| }, | |
| { | |
| "epoch": 1.339712918660287, | |
| "grad_norm": 0.6563542216109102, | |
| "learning_rate": 3.899640263455566e-05, | |
| "loss": 0.2349, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2351498007774353, | |
| "step": 840, | |
| "valid_targets_mean": 2515.1, | |
| "valid_targets_min": 1464 | |
| }, | |
| { | |
| "epoch": 1.3476874003189794, | |
| "grad_norm": 0.64783568591679, | |
| "learning_rate": 3.897137453184472e-05, | |
| "loss": 0.2372, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22898787260055542, | |
| "step": 845, | |
| "valid_targets_mean": 2141.3, | |
| "valid_targets_min": 596 | |
| }, | |
| { | |
| "epoch": 1.3556618819776713, | |
| "grad_norm": 0.5823873984785253, | |
| "learning_rate": 3.894604641356767e-05, | |
| "loss": 0.2323, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19752237200737, | |
| "step": 850, | |
| "valid_targets_mean": 2627.8, | |
| "valid_targets_min": 1357 | |
| }, | |
| { | |
| "epoch": 1.3636363636363638, | |
| "grad_norm": 0.5757894549438445, | |
| "learning_rate": 3.8920418680266346e-05, | |
| "loss": 0.2167, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1831761598587036, | |
| "step": 855, | |
| "valid_targets_mean": 2389.3, | |
| "valid_targets_min": 1272 | |
| }, | |
| { | |
| "epoch": 1.3716108452950557, | |
| "grad_norm": 0.49327863847117276, | |
| "learning_rate": 3.889449173722077e-05, | |
| "loss": 0.2172, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20380468666553497, | |
| "step": 860, | |
| "valid_targets_mean": 3471.6, | |
| "valid_targets_min": 1488 | |
| }, | |
| { | |
| "epoch": 1.3795853269537481, | |
| "grad_norm": 0.6307638441078381, | |
| "learning_rate": 3.8868265994442694e-05, | |
| "loss": 0.2014, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21134331822395325, | |
| "step": 865, | |
| "valid_targets_mean": 2447.6, | |
| "valid_targets_min": 947 | |
| }, | |
| { | |
| "epoch": 1.38755980861244, | |
| "grad_norm": 0.5265701690721363, | |
| "learning_rate": 3.8841741866669126e-05, | |
| "loss": 0.2454, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21514572203159332, | |
| "step": 870, | |
| "valid_targets_mean": 3132.0, | |
| "valid_targets_min": 753 | |
| }, | |
| { | |
| "epoch": 1.3955342902711323, | |
| "grad_norm": 0.6272683045446587, | |
| "learning_rate": 3.881491977335577e-05, | |
| "loss": 0.2304, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2270197570323944, | |
| "step": 875, | |
| "valid_targets_mean": 2979.2, | |
| "valid_targets_min": 1335 | |
| }, | |
| { | |
| "epoch": 1.4035087719298245, | |
| "grad_norm": 0.5125155204565588, | |
| "learning_rate": 3.878780013867038e-05, | |
| "loss": 0.2294, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.211686909198761, | |
| "step": 880, | |
| "valid_targets_mean": 3388.2, | |
| "valid_targets_min": 1407 | |
| }, | |
| { | |
| "epoch": 1.4114832535885167, | |
| "grad_norm": 0.5884010149005009, | |
| "learning_rate": 3.8760383391486074e-05, | |
| "loss": 0.2428, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21981844305992126, | |
| "step": 885, | |
| "valid_targets_mean": 2581.7, | |
| "valid_targets_min": 1090 | |
| }, | |
| { | |
| "epoch": 1.4194577352472089, | |
| "grad_norm": 0.5864469022550587, | |
| "learning_rate": 3.873266996537456e-05, | |
| "loss": 0.2394, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2843884825706482, | |
| "step": 890, | |
| "valid_targets_mean": 3282.1, | |
| "valid_targets_min": 1301 | |
| }, | |
| { | |
| "epoch": 1.427432216905901, | |
| "grad_norm": 0.6407016912115063, | |
| "learning_rate": 3.8704660298599225e-05, | |
| "loss": 0.2586, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3323836922645569, | |
| "step": 895, | |
| "valid_targets_mean": 3613.4, | |
| "valid_targets_min": 1151 | |
| }, | |
| { | |
| "epoch": 1.4354066985645932, | |
| "grad_norm": 0.548542676602486, | |
| "learning_rate": 3.867635483410827e-05, | |
| "loss": 0.2219, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2041669636964798, | |
| "step": 900, | |
| "valid_targets_mean": 2906.9, | |
| "valid_targets_min": 1046 | |
| }, | |
| { | |
| "epoch": 1.4433811802232854, | |
| "grad_norm": 0.5135362865237451, | |
| "learning_rate": 3.864775401952767e-05, | |
| "loss": 0.2504, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.236622616648674, | |
| "step": 905, | |
| "valid_targets_mean": 3590.2, | |
| "valid_targets_min": 1038 | |
| }, | |
| { | |
| "epoch": 1.4513556618819776, | |
| "grad_norm": 0.5695703284047577, | |
| "learning_rate": 3.8618858307154085e-05, | |
| "loss": 0.2381, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.31914830207824707, | |
| "step": 910, | |
| "valid_targets_mean": 3563.1, | |
| "valid_targets_min": 899 | |
| }, | |
| { | |
| "epoch": 1.4593301435406698, | |
| "grad_norm": 0.6066778005066594, | |
| "learning_rate": 3.8589668153947743e-05, | |
| "loss": 0.2182, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16905300319194794, | |
| "step": 915, | |
| "valid_targets_mean": 2114.1, | |
| "valid_targets_min": 1312 | |
| }, | |
| { | |
| "epoch": 1.467304625199362, | |
| "grad_norm": 0.5937089594035688, | |
| "learning_rate": 3.8560184021525194e-05, | |
| "loss": 0.2306, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2555321455001831, | |
| "step": 920, | |
| "valid_targets_mean": 2893.0, | |
| "valid_targets_min": 1271 | |
| }, | |
| { | |
| "epoch": 1.4752791068580542, | |
| "grad_norm": 0.6283464141673474, | |
| "learning_rate": 3.853040637615199e-05, | |
| "loss": 0.2278, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.195700004696846, | |
| "step": 925, | |
| "valid_targets_mean": 2522.1, | |
| "valid_targets_min": 1232 | |
| }, | |
| { | |
| "epoch": 1.4832535885167464, | |
| "grad_norm": 0.5723038985943404, | |
| "learning_rate": 3.850033568873536e-05, | |
| "loss": 0.2232, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19896413385868073, | |
| "step": 930, | |
| "valid_targets_mean": 2328.2, | |
| "valid_targets_min": 1076 | |
| }, | |
| { | |
| "epoch": 1.4912280701754386, | |
| "grad_norm": 0.6164522031974263, | |
| "learning_rate": 3.8469972434816706e-05, | |
| "loss": 0.2364, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18973922729492188, | |
| "step": 935, | |
| "valid_targets_mean": 2312.7, | |
| "valid_targets_min": 1280 | |
| }, | |
| { | |
| "epoch": 1.4992025518341308, | |
| "grad_norm": 0.6086637262441689, | |
| "learning_rate": 3.843931709456414e-05, | |
| "loss": 0.2323, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24724075198173523, | |
| "step": 940, | |
| "valid_targets_mean": 2995.4, | |
| "valid_targets_min": 1567 | |
| }, | |
| { | |
| "epoch": 1.507177033492823, | |
| "grad_norm": 0.6483381919678638, | |
| "learning_rate": 3.840837015276483e-05, | |
| "loss": 0.2483, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3337307572364807, | |
| "step": 945, | |
| "valid_targets_mean": 2911.9, | |
| "valid_targets_min": 1306 | |
| }, | |
| { | |
| "epoch": 1.5151515151515151, | |
| "grad_norm": 0.5855323586411422, | |
| "learning_rate": 3.83771320988174e-05, | |
| "loss": 0.2305, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17903995513916016, | |
| "step": 950, | |
| "valid_targets_mean": 2291.5, | |
| "valid_targets_min": 1377 | |
| }, | |
| { | |
| "epoch": 1.5231259968102073, | |
| "grad_norm": 0.584270941139963, | |
| "learning_rate": 3.834560342672413e-05, | |
| "loss": 0.2357, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21279588341712952, | |
| "step": 955, | |
| "valid_targets_mean": 2610.5, | |
| "valid_targets_min": 1268 | |
| }, | |
| { | |
| "epoch": 1.5311004784688995, | |
| "grad_norm": 0.5503620072606771, | |
| "learning_rate": 3.831378463508318e-05, | |
| "loss": 0.222, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2559887170791626, | |
| "step": 960, | |
| "valid_targets_mean": 3138.2, | |
| "valid_targets_min": 1078 | |
| }, | |
| { | |
| "epoch": 1.5390749601275917, | |
| "grad_norm": 0.6001392975485096, | |
| "learning_rate": 3.8281676227080694e-05, | |
| "loss": 0.2222, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2329024225473404, | |
| "step": 965, | |
| "valid_targets_mean": 2648.9, | |
| "valid_targets_min": 1190 | |
| }, | |
| { | |
| "epoch": 1.547049441786284, | |
| "grad_norm": 0.5346724503885805, | |
| "learning_rate": 3.824927871048284e-05, | |
| "loss": 0.2146, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19294703006744385, | |
| "step": 970, | |
| "valid_targets_mean": 2822.2, | |
| "valid_targets_min": 1261 | |
| }, | |
| { | |
| "epoch": 1.555023923444976, | |
| "grad_norm": 0.49919442227666905, | |
| "learning_rate": 3.8216592597627797e-05, | |
| "loss": 0.2022, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2007545828819275, | |
| "step": 975, | |
| "valid_targets_mean": 3116.3, | |
| "valid_targets_min": 1066 | |
| }, | |
| { | |
| "epoch": 1.5629984051036683, | |
| "grad_norm": 0.5982205761118605, | |
| "learning_rate": 3.818361840541761e-05, | |
| "loss": 0.2098, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18389758467674255, | |
| "step": 980, | |
| "valid_targets_mean": 2047.2, | |
| "valid_targets_min": 959 | |
| }, | |
| { | |
| "epoch": 1.5709728867623605, | |
| "grad_norm": 0.5512917759429814, | |
| "learning_rate": 3.815035665531008e-05, | |
| "loss": 0.2115, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1839146614074707, | |
| "step": 985, | |
| "valid_targets_mean": 2353.8, | |
| "valid_targets_min": 684 | |
| }, | |
| { | |
| "epoch": 1.5789473684210527, | |
| "grad_norm": 0.6223649318682416, | |
| "learning_rate": 3.811680787331047e-05, | |
| "loss": 0.2182, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18036724627017975, | |
| "step": 990, | |
| "valid_targets_mean": 2156.6, | |
| "valid_targets_min": 1002 | |
| }, | |
| { | |
| "epoch": 1.5869218500797448, | |
| "grad_norm": 0.6352872088960317, | |
| "learning_rate": 3.8082972589963175e-05, | |
| "loss": 0.2359, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2819938361644745, | |
| "step": 995, | |
| "valid_targets_mean": 2302.4, | |
| "valid_targets_min": 894 | |
| }, | |
| { | |
| "epoch": 1.594896331738437, | |
| "grad_norm": 0.6242377091078459, | |
| "learning_rate": 3.80488513403434e-05, | |
| "loss": 0.2396, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.289346307516098, | |
| "step": 1000, | |
| "valid_targets_mean": 2421.4, | |
| "valid_targets_min": 857 | |
| }, | |
| { | |
| "epoch": 1.6028708133971292, | |
| "grad_norm": 0.5826983613530846, | |
| "learning_rate": 3.8014444664048616e-05, | |
| "loss": 0.2144, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25043565034866333, | |
| "step": 1005, | |
| "valid_targets_mean": 2823.9, | |
| "valid_targets_min": 1114 | |
| }, | |
| { | |
| "epoch": 1.6108452950558214, | |
| "grad_norm": 0.5546186494842212, | |
| "learning_rate": 3.797975310519009e-05, | |
| "loss": 0.2145, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21974100172519684, | |
| "step": 1010, | |
| "valid_targets_mean": 3072.6, | |
| "valid_targets_min": 1653 | |
| }, | |
| { | |
| "epoch": 1.6188197767145136, | |
| "grad_norm": 0.4632694182062778, | |
| "learning_rate": 3.794477721238425e-05, | |
| "loss": 0.2261, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21472452580928802, | |
| "step": 1015, | |
| "valid_targets_mean": 3615.5, | |
| "valid_targets_min": 1532 | |
| }, | |
| { | |
| "epoch": 1.6267942583732058, | |
| "grad_norm": 0.5830152765455087, | |
| "learning_rate": 3.7909517538744e-05, | |
| "loss": 0.2408, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2194686233997345, | |
| "step": 1020, | |
| "valid_targets_mean": 2642.8, | |
| "valid_targets_min": 1392 | |
| }, | |
| { | |
| "epoch": 1.6347687400318978, | |
| "grad_norm": 0.6151873688159493, | |
| "learning_rate": 3.7873974641870006e-05, | |
| "loss": 0.2328, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1882760226726532, | |
| "step": 1025, | |
| "valid_targets_mean": 2297.1, | |
| "valid_targets_min": 1269 | |
| }, | |
| { | |
| "epoch": 1.6427432216905902, | |
| "grad_norm": 2.140862445980475, | |
| "learning_rate": 3.7838149083841856e-05, | |
| "loss": 0.227, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23605549335479736, | |
| "step": 1030, | |
| "valid_targets_mean": 2535.2, | |
| "valid_targets_min": 1069 | |
| }, | |
| { | |
| "epoch": 1.6507177033492821, | |
| "grad_norm": 0.6528803837865146, | |
| "learning_rate": 3.7802041431209166e-05, | |
| "loss": 0.2556, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2378740757703781, | |
| "step": 1035, | |
| "valid_targets_mean": 2299.6, | |
| "valid_targets_min": 743 | |
| }, | |
| { | |
| "epoch": 1.6586921850079746, | |
| "grad_norm": 0.6191369201573628, | |
| "learning_rate": 3.776565225498264e-05, | |
| "loss": 0.2877, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24140696227550507, | |
| "step": 1040, | |
| "valid_targets_mean": 2222.6, | |
| "valid_targets_min": 1094 | |
| }, | |
| { | |
| "epoch": 1.6666666666666665, | |
| "grad_norm": 0.6498039390645324, | |
| "learning_rate": 3.7728982130625025e-05, | |
| "loss": 0.2179, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2488032430410385, | |
| "step": 1045, | |
| "valid_targets_mean": 2576.2, | |
| "valid_targets_min": 1378 | |
| }, | |
| { | |
| "epoch": 1.674641148325359, | |
| "grad_norm": 0.6375130250679868, | |
| "learning_rate": 3.769203163804202e-05, | |
| "loss": 0.2556, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2386590838432312, | |
| "step": 1050, | |
| "valid_targets_mean": 2930.1, | |
| "valid_targets_min": 1121 | |
| }, | |
| { | |
| "epoch": 1.682615629984051, | |
| "grad_norm": 0.5527652926140382, | |
| "learning_rate": 3.7654801361573076e-05, | |
| "loss": 0.2367, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.264995276927948, | |
| "step": 1055, | |
| "valid_targets_mean": 3239.4, | |
| "valid_targets_min": 1355 | |
| }, | |
| { | |
| "epoch": 1.6905901116427433, | |
| "grad_norm": 0.5370865034755017, | |
| "learning_rate": 3.761729188998222e-05, | |
| "loss": 0.2418, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.28277504444122314, | |
| "step": 1060, | |
| "valid_targets_mean": 3348.9, | |
| "valid_targets_min": 876 | |
| }, | |
| { | |
| "epoch": 1.6985645933014353, | |
| "grad_norm": 0.5747670458007346, | |
| "learning_rate": 3.757950381644868e-05, | |
| "loss": 0.2282, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20245248079299927, | |
| "step": 1065, | |
| "valid_targets_mean": 2440.8, | |
| "valid_targets_min": 1309 | |
| }, | |
| { | |
| "epoch": 1.7065390749601277, | |
| "grad_norm": 0.6094654650302161, | |
| "learning_rate": 3.7541437738557524e-05, | |
| "loss": 0.2247, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20767079293727875, | |
| "step": 1070, | |
| "valid_targets_mean": 2374.6, | |
| "valid_targets_min": 1089 | |
| }, | |
| { | |
| "epoch": 1.7145135566188197, | |
| "grad_norm": 0.5569861380782473, | |
| "learning_rate": 3.750309425829022e-05, | |
| "loss": 0.2408, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24564585089683533, | |
| "step": 1075, | |
| "valid_targets_mean": 3401.3, | |
| "valid_targets_min": 1262 | |
| }, | |
| { | |
| "epoch": 1.722488038277512, | |
| "grad_norm": 0.613745920593988, | |
| "learning_rate": 3.746447398201512e-05, | |
| "loss": 0.2132, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2049260288476944, | |
| "step": 1080, | |
| "valid_targets_mean": 2619.6, | |
| "valid_targets_min": 1067 | |
| }, | |
| { | |
| "epoch": 1.730462519936204, | |
| "grad_norm": 0.5548189655462026, | |
| "learning_rate": 3.7425577520477846e-05, | |
| "loss": 0.2516, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23266099393367767, | |
| "step": 1085, | |
| "valid_targets_mean": 2925.9, | |
| "valid_targets_min": 1192 | |
| }, | |
| { | |
| "epoch": 1.7384370015948964, | |
| "grad_norm": 0.5254275868939079, | |
| "learning_rate": 3.738640548879166e-05, | |
| "loss": 0.2263, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18870019912719727, | |
| "step": 1090, | |
| "valid_targets_mean": 2759.0, | |
| "valid_targets_min": 1227 | |
| }, | |
| { | |
| "epoch": 1.7464114832535884, | |
| "grad_norm": 0.5020906772682225, | |
| "learning_rate": 3.7346958506427696e-05, | |
| "loss": 0.2434, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25047755241394043, | |
| "step": 1095, | |
| "valid_targets_mean": 3543.0, | |
| "valid_targets_min": 1238 | |
| }, | |
| { | |
| "epoch": 1.7543859649122808, | |
| "grad_norm": 0.6019040221940559, | |
| "learning_rate": 3.730723719720523e-05, | |
| "loss": 0.2188, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21497926115989685, | |
| "step": 1100, | |
| "valid_targets_mean": 2993.9, | |
| "valid_targets_min": 1150 | |
| }, | |
| { | |
| "epoch": 1.7623604465709728, | |
| "grad_norm": 0.5134617499391168, | |
| "learning_rate": 3.7267242189281746e-05, | |
| "loss": 0.217, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23121070861816406, | |
| "step": 1105, | |
| "valid_targets_mean": 3359.8, | |
| "valid_targets_min": 1015 | |
| }, | |
| { | |
| "epoch": 1.7703349282296652, | |
| "grad_norm": 0.555534809215992, | |
| "learning_rate": 3.722697411514305e-05, | |
| "loss": 0.223, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2400152087211609, | |
| "step": 1110, | |
| "valid_targets_mean": 2866.0, | |
| "valid_targets_min": 1332 | |
| }, | |
| { | |
| "epoch": 1.7783094098883572, | |
| "grad_norm": 0.6105348331308175, | |
| "learning_rate": 3.7186433611593225e-05, | |
| "loss": 0.2265, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.30984336137771606, | |
| "step": 1115, | |
| "valid_targets_mean": 3165.9, | |
| "valid_targets_min": 946 | |
| }, | |
| { | |
| "epoch": 1.7862838915470496, | |
| "grad_norm": 0.5761430256224706, | |
| "learning_rate": 3.7145621319744614e-05, | |
| "loss": 0.2334, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26511406898498535, | |
| "step": 1120, | |
| "valid_targets_mean": 3365.7, | |
| "valid_targets_min": 840 | |
| }, | |
| { | |
| "epoch": 1.7942583732057416, | |
| "grad_norm": 0.5782063968492165, | |
| "learning_rate": 3.7104537885007635e-05, | |
| "loss": 0.2262, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21325793862342834, | |
| "step": 1125, | |
| "valid_targets_mean": 2586.2, | |
| "valid_targets_min": 1150 | |
| }, | |
| { | |
| "epoch": 1.802232854864434, | |
| "grad_norm": 0.6766750855241791, | |
| "learning_rate": 3.7063183957080594e-05, | |
| "loss": 0.2319, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19376319646835327, | |
| "step": 1130, | |
| "valid_targets_mean": 2382.9, | |
| "valid_targets_min": 1198 | |
| }, | |
| { | |
| "epoch": 1.810207336523126, | |
| "grad_norm": 0.47421899311917426, | |
| "learning_rate": 3.7021560189939416e-05, | |
| "loss": 0.2425, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21126791834831238, | |
| "step": 1135, | |
| "valid_targets_mean": 3506.8, | |
| "valid_targets_min": 1255 | |
| }, | |
| { | |
| "epoch": 1.8181818181818183, | |
| "grad_norm": 0.5034779048358913, | |
| "learning_rate": 3.697966724182729e-05, | |
| "loss": 0.2013, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19855384528636932, | |
| "step": 1140, | |
| "valid_targets_mean": 3456.3, | |
| "valid_targets_min": 1199 | |
| }, | |
| { | |
| "epoch": 1.8261562998405103, | |
| "grad_norm": 0.6337585629876568, | |
| "learning_rate": 3.6937505775244246e-05, | |
| "loss": 0.2078, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22304585576057434, | |
| "step": 1145, | |
| "valid_targets_mean": 2685.5, | |
| "valid_targets_min": 1248 | |
| }, | |
| { | |
| "epoch": 1.8341307814992025, | |
| "grad_norm": 0.5610178719430047, | |
| "learning_rate": 3.689507645693674e-05, | |
| "loss": 0.2166, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2178836166858673, | |
| "step": 1150, | |
| "valid_targets_mean": 3309.5, | |
| "valid_targets_min": 1342 | |
| }, | |
| { | |
| "epoch": 1.8421052631578947, | |
| "grad_norm": 0.5239047956067104, | |
| "learning_rate": 3.6852379957887025e-05, | |
| "loss": 0.2129, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1819612830877304, | |
| "step": 1155, | |
| "valid_targets_mean": 2517.4, | |
| "valid_targets_min": 1369 | |
| }, | |
| { | |
| "epoch": 1.8500797448165869, | |
| "grad_norm": 0.5749497640819529, | |
| "learning_rate": 3.6809416953302606e-05, | |
| "loss": 0.2269, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1893683671951294, | |
| "step": 1160, | |
| "valid_targets_mean": 2465.4, | |
| "valid_targets_min": 1262 | |
| }, | |
| { | |
| "epoch": 1.858054226475279, | |
| "grad_norm": 0.5625715315011639, | |
| "learning_rate": 3.676618812260553e-05, | |
| "loss": 0.2083, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18896302580833435, | |
| "step": 1165, | |
| "valid_targets_mean": 2504.8, | |
| "valid_targets_min": 1078 | |
| }, | |
| { | |
| "epoch": 1.8660287081339713, | |
| "grad_norm": 0.5196211344002306, | |
| "learning_rate": 3.672269414942166e-05, | |
| "loss": 0.2349, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18706221878528595, | |
| "step": 1170, | |
| "valid_targets_mean": 2709.4, | |
| "valid_targets_min": 1589 | |
| }, | |
| { | |
| "epoch": 1.8740031897926634, | |
| "grad_norm": 0.5409333832286379, | |
| "learning_rate": 3.6678935721569825e-05, | |
| "loss": 0.2275, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18008098006248474, | |
| "step": 1175, | |
| "valid_targets_mean": 2698.2, | |
| "valid_targets_min": 1756 | |
| }, | |
| { | |
| "epoch": 1.8819776714513556, | |
| "grad_norm": 0.5215242217921682, | |
| "learning_rate": 3.663491353105101e-05, | |
| "loss": 0.2165, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23432385921478271, | |
| "step": 1180, | |
| "valid_targets_mean": 3169.7, | |
| "valid_targets_min": 985 | |
| }, | |
| { | |
| "epoch": 1.8899521531100478, | |
| "grad_norm": 0.4659554830522906, | |
| "learning_rate": 3.659062827403735e-05, | |
| "loss": 0.1949, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18295270204544067, | |
| "step": 1185, | |
| "valid_targets_mean": 3136.5, | |
| "valid_targets_min": 878 | |
| }, | |
| { | |
| "epoch": 1.89792663476874, | |
| "grad_norm": 0.5834321903658111, | |
| "learning_rate": 3.654608065086115e-05, | |
| "loss": 0.1986, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1885906457901001, | |
| "step": 1190, | |
| "valid_targets_mean": 2347.2, | |
| "valid_targets_min": 1357 | |
| }, | |
| { | |
| "epoch": 1.9059011164274322, | |
| "grad_norm": 0.4892253621146179, | |
| "learning_rate": 3.650127136600379e-05, | |
| "loss": 0.2282, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25364458560943604, | |
| "step": 1195, | |
| "valid_targets_mean": 4085.2, | |
| "valid_targets_min": 1349 | |
| }, | |
| { | |
| "epoch": 1.9138755980861244, | |
| "grad_norm": 0.5076087679251333, | |
| "learning_rate": 3.645620112808464e-05, | |
| "loss": 0.216, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20003008842468262, | |
| "step": 1200, | |
| "valid_targets_mean": 3066.7, | |
| "valid_targets_min": 1299 | |
| }, | |
| { | |
| "epoch": 1.9218500797448166, | |
| "grad_norm": 0.5335946413112277, | |
| "learning_rate": 3.641087064984977e-05, | |
| "loss": 0.2257, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2127196490764618, | |
| "step": 1205, | |
| "valid_targets_mean": 2720.1, | |
| "valid_targets_min": 789 | |
| }, | |
| { | |
| "epoch": 1.9298245614035088, | |
| "grad_norm": 0.6361693621882534, | |
| "learning_rate": 3.636528064816073e-05, | |
| "loss": 0.2213, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19323310256004333, | |
| "step": 1210, | |
| "valid_targets_mean": 2129.1, | |
| "valid_targets_min": 904 | |
| }, | |
| { | |
| "epoch": 1.937799043062201, | |
| "grad_norm": 0.5843343837357229, | |
| "learning_rate": 3.6319431843983223e-05, | |
| "loss": 0.2507, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2852637767791748, | |
| "step": 1215, | |
| "valid_targets_mean": 3620.1, | |
| "valid_targets_min": 1158 | |
| }, | |
| { | |
| "epoch": 1.9457735247208932, | |
| "grad_norm": 0.5951676671617808, | |
| "learning_rate": 3.6273324962375676e-05, | |
| "loss": 0.2073, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2468632161617279, | |
| "step": 1220, | |
| "valid_targets_mean": 2911.6, | |
| "valid_targets_min": 1086 | |
| }, | |
| { | |
| "epoch": 1.9537480063795853, | |
| "grad_norm": 0.6569118735077601, | |
| "learning_rate": 3.622696073247777e-05, | |
| "loss": 0.2117, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24635280668735504, | |
| "step": 1225, | |
| "valid_targets_mean": 3458.5, | |
| "valid_targets_min": 692 | |
| }, | |
| { | |
| "epoch": 1.9617224880382775, | |
| "grad_norm": 0.5437346633446027, | |
| "learning_rate": 3.6180339887498953e-05, | |
| "loss": 0.2272, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2951074540615082, | |
| "step": 1230, | |
| "valid_targets_mean": 3982.8, | |
| "valid_targets_min": 1143 | |
| }, | |
| { | |
| "epoch": 1.9696969696969697, | |
| "grad_norm": 0.4940449249242946, | |
| "learning_rate": 3.613346316470678e-05, | |
| "loss": 0.2376, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23175011575222015, | |
| "step": 1235, | |
| "valid_targets_mean": 3729.4, | |
| "valid_targets_min": 1226 | |
| }, | |
| { | |
| "epoch": 1.977671451355662, | |
| "grad_norm": 0.5477246251152017, | |
| "learning_rate": 3.60863313054153e-05, | |
| "loss": 0.215, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18664443492889404, | |
| "step": 1240, | |
| "valid_targets_mean": 2555.5, | |
| "valid_targets_min": 1347 | |
| }, | |
| { | |
| "epoch": 1.985645933014354, | |
| "grad_norm": 0.5424225434810162, | |
| "learning_rate": 3.6038945054973334e-05, | |
| "loss": 0.2151, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21529079973697662, | |
| "step": 1245, | |
| "valid_targets_mean": 3182.9, | |
| "valid_targets_min": 1049 | |
| }, | |
| { | |
| "epoch": 1.9936204146730463, | |
| "grad_norm": 0.5309854104241837, | |
| "learning_rate": 3.599130516275266e-05, | |
| "loss": 0.2126, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21649429202079773, | |
| "step": 1250, | |
| "valid_targets_mean": 2814.7, | |
| "valid_targets_min": 976 | |
| }, | |
| { | |
| "epoch": 2.0015948963317385, | |
| "grad_norm": 0.6830016071881365, | |
| "learning_rate": 3.594341238213618e-05, | |
| "loss": 0.2155, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2428247332572937, | |
| "step": 1255, | |
| "valid_targets_mean": 2230.6, | |
| "valid_targets_min": 1201 | |
| }, | |
| { | |
| "epoch": 2.0095693779904304, | |
| "grad_norm": 0.583613177123442, | |
| "learning_rate": 3.589526747050601e-05, | |
| "loss": 0.1952, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1945180892944336, | |
| "step": 1260, | |
| "valid_targets_mean": 2468.8, | |
| "valid_targets_min": 1106 | |
| }, | |
| { | |
| "epoch": 2.017543859649123, | |
| "grad_norm": 0.5249288021444334, | |
| "learning_rate": 3.584687118923149e-05, | |
| "loss": 0.1905, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1898598074913025, | |
| "step": 1265, | |
| "valid_targets_mean": 2771.9, | |
| "valid_targets_min": 507 | |
| }, | |
| { | |
| "epoch": 2.025518341307815, | |
| "grad_norm": 0.6525341636876454, | |
| "learning_rate": 3.579822430365714e-05, | |
| "loss": 0.1966, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22228668630123138, | |
| "step": 1270, | |
| "valid_targets_mean": 3191.8, | |
| "valid_targets_min": 1374 | |
| }, | |
| { | |
| "epoch": 2.0334928229665072, | |
| "grad_norm": 0.5729538709639772, | |
| "learning_rate": 3.57493275830906e-05, | |
| "loss": 0.1909, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17011548578739166, | |
| "step": 1275, | |
| "valid_targets_mean": 2719.9, | |
| "valid_targets_min": 1164 | |
| }, | |
| { | |
| "epoch": 2.041467304625199, | |
| "grad_norm": 0.5358768215183619, | |
| "learning_rate": 3.570018180079037e-05, | |
| "loss": 0.185, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21794959902763367, | |
| "step": 1280, | |
| "valid_targets_mean": 3458.1, | |
| "valid_targets_min": 1452 | |
| }, | |
| { | |
| "epoch": 2.0494417862838916, | |
| "grad_norm": 1.025831647271483, | |
| "learning_rate": 3.5650787733953715e-05, | |
| "loss": 0.1814, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16724932193756104, | |
| "step": 1285, | |
| "valid_targets_mean": 2740.6, | |
| "valid_targets_min": 1379 | |
| }, | |
| { | |
| "epoch": 2.0574162679425836, | |
| "grad_norm": 0.5249612579847214, | |
| "learning_rate": 3.560114616370425e-05, | |
| "loss": 0.222, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21160989999771118, | |
| "step": 1290, | |
| "valid_targets_mean": 3251.8, | |
| "valid_targets_min": 1721 | |
| }, | |
| { | |
| "epoch": 2.065390749601276, | |
| "grad_norm": 0.5644105309104152, | |
| "learning_rate": 3.555125787507964e-05, | |
| "loss": 0.1868, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19360047578811646, | |
| "step": 1295, | |
| "valid_targets_mean": 2911.9, | |
| "valid_targets_min": 959 | |
| }, | |
| { | |
| "epoch": 2.073365231259968, | |
| "grad_norm": 0.5815407495575634, | |
| "learning_rate": 3.550112365701921e-05, | |
| "loss": 0.2272, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2966446280479431, | |
| "step": 1300, | |
| "valid_targets_mean": 3293.7, | |
| "valid_targets_min": 789 | |
| }, | |
| { | |
| "epoch": 2.0813397129186604, | |
| "grad_norm": 0.5385897882404551, | |
| "learning_rate": 3.545074430235142e-05, | |
| "loss": 0.21, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22409307956695557, | |
| "step": 1305, | |
| "valid_targets_mean": 3228.1, | |
| "valid_targets_min": 1089 | |
| }, | |
| { | |
| "epoch": 2.0893141945773523, | |
| "grad_norm": 0.5692012223049029, | |
| "learning_rate": 3.540012060778137e-05, | |
| "loss": 0.18, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15392714738845825, | |
| "step": 1310, | |
| "valid_targets_mean": 2260.0, | |
| "valid_targets_min": 1031 | |
| }, | |
| { | |
| "epoch": 2.0972886762360448, | |
| "grad_norm": 0.5300644084507816, | |
| "learning_rate": 3.534925337387816e-05, | |
| "loss": 0.2093, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24721625447273254, | |
| "step": 1315, | |
| "valid_targets_mean": 3622.4, | |
| "valid_targets_min": 1160 | |
| }, | |
| { | |
| "epoch": 2.1052631578947367, | |
| "grad_norm": 0.5647284692712146, | |
| "learning_rate": 3.529814340506226e-05, | |
| "loss": 0.2017, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20379406213760376, | |
| "step": 1320, | |
| "valid_targets_mean": 3280.3, | |
| "valid_targets_min": 1210 | |
| }, | |
| { | |
| "epoch": 2.113237639553429, | |
| "grad_norm": 0.558965345583107, | |
| "learning_rate": 3.524679150959277e-05, | |
| "loss": 0.2158, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17460405826568604, | |
| "step": 1325, | |
| "valid_targets_mean": 2684.2, | |
| "valid_targets_min": 1067 | |
| }, | |
| { | |
| "epoch": 2.121212121212121, | |
| "grad_norm": 0.5955047163911827, | |
| "learning_rate": 3.519519849955466e-05, | |
| "loss": 0.2073, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1885492503643036, | |
| "step": 1330, | |
| "valid_targets_mean": 3299.5, | |
| "valid_targets_min": 1561 | |
| }, | |
| { | |
| "epoch": 2.1291866028708135, | |
| "grad_norm": 0.4966909790003643, | |
| "learning_rate": 3.514336519084591e-05, | |
| "loss": 0.1976, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18052926659584045, | |
| "step": 1335, | |
| "valid_targets_mean": 3317.4, | |
| "valid_targets_min": 1262 | |
| }, | |
| { | |
| "epoch": 2.1371610845295055, | |
| "grad_norm": 0.5299411406686442, | |
| "learning_rate": 3.509129240316461e-05, | |
| "loss": 0.178, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2125854790210724, | |
| "step": 1340, | |
| "valid_targets_mean": 3120.4, | |
| "valid_targets_min": 1194 | |
| }, | |
| { | |
| "epoch": 2.145135566188198, | |
| "grad_norm": 0.5800308757617607, | |
| "learning_rate": 3.5038980959995985e-05, | |
| "loss": 0.2093, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.263176828622818, | |
| "step": 1345, | |
| "valid_targets_mean": 3585.3, | |
| "valid_targets_min": 1159 | |
| }, | |
| { | |
| "epoch": 2.15311004784689, | |
| "grad_norm": 0.5999009997718916, | |
| "learning_rate": 3.498643168859941e-05, | |
| "loss": 0.2003, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20120833814144135, | |
| "step": 1350, | |
| "valid_targets_mean": 2516.6, | |
| "valid_targets_min": 946 | |
| }, | |
| { | |
| "epoch": 2.1610845295055823, | |
| "grad_norm": 0.48961821808382117, | |
| "learning_rate": 3.493364541999529e-05, | |
| "loss": 0.2146, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18192213773727417, | |
| "step": 1355, | |
| "valid_targets_mean": 3155.0, | |
| "valid_targets_min": 1281 | |
| }, | |
| { | |
| "epoch": 2.1690590111642742, | |
| "grad_norm": 0.6849100630435138, | |
| "learning_rate": 3.488062298895194e-05, | |
| "loss": 0.1911, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18514913320541382, | |
| "step": 1360, | |
| "valid_targets_mean": 2309.5, | |
| "valid_targets_min": 894 | |
| }, | |
| { | |
| "epoch": 2.1770334928229667, | |
| "grad_norm": 0.5344472039018315, | |
| "learning_rate": 3.482736523397237e-05, | |
| "loss": 0.1968, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1777096837759018, | |
| "step": 1365, | |
| "valid_targets_mean": 2620.4, | |
| "valid_targets_min": 1172 | |
| }, | |
| { | |
| "epoch": 2.1850079744816586, | |
| "grad_norm": 0.5312536211636795, | |
| "learning_rate": 3.4773872997281026e-05, | |
| "loss": 0.2187, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2192026972770691, | |
| "step": 1370, | |
| "valid_targets_mean": 3280.9, | |
| "valid_targets_min": 1436 | |
| }, | |
| { | |
| "epoch": 2.192982456140351, | |
| "grad_norm": 0.5286821474310076, | |
| "learning_rate": 3.472014712481048e-05, | |
| "loss": 0.2016, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19956275820732117, | |
| "step": 1375, | |
| "valid_targets_mean": 3385.3, | |
| "valid_targets_min": 812 | |
| }, | |
| { | |
| "epoch": 2.200956937799043, | |
| "grad_norm": 0.5500060285804033, | |
| "learning_rate": 3.466618846618806e-05, | |
| "loss": 0.2139, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18642191588878632, | |
| "step": 1380, | |
| "valid_targets_mean": 2684.4, | |
| "valid_targets_min": 1302 | |
| }, | |
| { | |
| "epoch": 2.2089314194577354, | |
| "grad_norm": 0.5352427652804307, | |
| "learning_rate": 3.461199787472238e-05, | |
| "loss": 0.1931, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18637782335281372, | |
| "step": 1385, | |
| "valid_targets_mean": 3044.4, | |
| "valid_targets_min": 986 | |
| }, | |
| { | |
| "epoch": 2.2169059011164274, | |
| "grad_norm": 0.5447384914029251, | |
| "learning_rate": 3.455757620738989e-05, | |
| "loss": 0.1974, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18728208541870117, | |
| "step": 1390, | |
| "valid_targets_mean": 3003.2, | |
| "valid_targets_min": 1236 | |
| }, | |
| { | |
| "epoch": 2.22488038277512, | |
| "grad_norm": 0.5717238697325798, | |
| "learning_rate": 3.450292432482127e-05, | |
| "loss": 0.2048, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2106652557849884, | |
| "step": 1395, | |
| "valid_targets_mean": 2633.9, | |
| "valid_targets_min": 1250 | |
| }, | |
| { | |
| "epoch": 2.2328548644338118, | |
| "grad_norm": 0.6406144106784887, | |
| "learning_rate": 3.444804309128789e-05, | |
| "loss": 0.2116, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2285759598016739, | |
| "step": 1400, | |
| "valid_targets_mean": 3575.9, | |
| "valid_targets_min": 1128 | |
| }, | |
| { | |
| "epoch": 2.240829346092504, | |
| "grad_norm": 0.47144056813631574, | |
| "learning_rate": 3.439293337468808e-05, | |
| "loss": 0.1921, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1521637737751007, | |
| "step": 1405, | |
| "valid_targets_mean": 2935.6, | |
| "valid_targets_min": 1082 | |
| }, | |
| { | |
| "epoch": 2.248803827751196, | |
| "grad_norm": 0.5527496205056719, | |
| "learning_rate": 3.4337596046533426e-05, | |
| "loss": 0.2284, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2488706111907959, | |
| "step": 1410, | |
| "valid_targets_mean": 3477.2, | |
| "valid_targets_min": 1027 | |
| }, | |
| { | |
| "epoch": 2.2567783094098885, | |
| "grad_norm": 0.508828813177984, | |
| "learning_rate": 3.4282031981935e-05, | |
| "loss": 0.2176, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2234359085559845, | |
| "step": 1415, | |
| "valid_targets_mean": 3965.8, | |
| "valid_targets_min": 1125 | |
| }, | |
| { | |
| "epoch": 2.2647527910685805, | |
| "grad_norm": 0.5695739787713568, | |
| "learning_rate": 3.42262420595895e-05, | |
| "loss": 0.1895, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24366150796413422, | |
| "step": 1420, | |
| "valid_targets_mean": 3077.3, | |
| "valid_targets_min": 1028 | |
| }, | |
| { | |
| "epoch": 2.2727272727272725, | |
| "grad_norm": 0.5998788145463537, | |
| "learning_rate": 3.417022716176539e-05, | |
| "loss": 0.2166, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2406616061925888, | |
| "step": 1425, | |
| "valid_targets_mean": 2459.1, | |
| "valid_targets_min": 1059 | |
| }, | |
| { | |
| "epoch": 2.280701754385965, | |
| "grad_norm": 0.5743859087966089, | |
| "learning_rate": 3.411398817428889e-05, | |
| "loss": 0.1926, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17860084772109985, | |
| "step": 1430, | |
| "valid_targets_mean": 2430.9, | |
| "valid_targets_min": 1288 | |
| }, | |
| { | |
| "epoch": 2.2886762360446573, | |
| "grad_norm": 0.5291060692515646, | |
| "learning_rate": 3.4057525986530016e-05, | |
| "loss": 0.1989, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2760910391807556, | |
| "step": 1435, | |
| "valid_targets_mean": 4081.8, | |
| "valid_targets_min": 1333 | |
| }, | |
| { | |
| "epoch": 2.2966507177033493, | |
| "grad_norm": 0.4928425236453258, | |
| "learning_rate": 3.400084149138851e-05, | |
| "loss": 0.2057, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20123213529586792, | |
| "step": 1440, | |
| "valid_targets_mean": 3745.3, | |
| "valid_targets_min": 1271 | |
| }, | |
| { | |
| "epoch": 2.3046251993620412, | |
| "grad_norm": 0.5186264077191272, | |
| "learning_rate": 3.394393558527969e-05, | |
| "loss": 0.1822, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1761550009250641, | |
| "step": 1445, | |
| "valid_targets_mean": 3007.8, | |
| "valid_targets_min": 1280 | |
| }, | |
| { | |
| "epoch": 2.3125996810207337, | |
| "grad_norm": 0.587444586310813, | |
| "learning_rate": 3.388680916812031e-05, | |
| "loss": 0.1857, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16212230920791626, | |
| "step": 1450, | |
| "valid_targets_mean": 2444.2, | |
| "valid_targets_min": 692 | |
| }, | |
| { | |
| "epoch": 2.320574162679426, | |
| "grad_norm": 0.5679214383824454, | |
| "learning_rate": 3.382946314331429e-05, | |
| "loss": 0.2167, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2232741117477417, | |
| "step": 1455, | |
| "valid_targets_mean": 2940.7, | |
| "valid_targets_min": 991 | |
| }, | |
| { | |
| "epoch": 2.328548644338118, | |
| "grad_norm": 0.7826501249243136, | |
| "learning_rate": 3.377189841773848e-05, | |
| "loss": 0.1887, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.25218990445137024, | |
| "step": 1460, | |
| "valid_targets_mean": 2453.1, | |
| "valid_targets_min": 1053 | |
| }, | |
| { | |
| "epoch": 2.33652312599681, | |
| "grad_norm": 0.5876238136730855, | |
| "learning_rate": 3.371411590172827e-05, | |
| "loss": 0.199, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17968472838401794, | |
| "step": 1465, | |
| "valid_targets_mean": 2336.6, | |
| "valid_targets_min": 1328 | |
| }, | |
| { | |
| "epoch": 2.3444976076555024, | |
| "grad_norm": 0.8075832611828215, | |
| "learning_rate": 3.365611650906321e-05, | |
| "loss": 0.1866, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1817243993282318, | |
| "step": 1470, | |
| "valid_targets_mean": 2379.7, | |
| "valid_targets_min": 1116 | |
| }, | |
| { | |
| "epoch": 2.352472089314195, | |
| "grad_norm": 0.5509881788274565, | |
| "learning_rate": 3.359790115695259e-05, | |
| "loss": 0.2206, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22982379794120789, | |
| "step": 1475, | |
| "valid_targets_mean": 3067.1, | |
| "valid_targets_min": 993 | |
| }, | |
| { | |
| "epoch": 2.360446570972887, | |
| "grad_norm": 0.5056847617580823, | |
| "learning_rate": 3.353947076602088e-05, | |
| "loss": 0.1747, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1958909034729004, | |
| "step": 1480, | |
| "valid_targets_mean": 3523.2, | |
| "valid_targets_min": 1121 | |
| }, | |
| { | |
| "epoch": 2.3684210526315788, | |
| "grad_norm": 0.5159845635621267, | |
| "learning_rate": 3.34808262602932e-05, | |
| "loss": 0.1831, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16456438601016998, | |
| "step": 1485, | |
| "valid_targets_mean": 3085.1, | |
| "valid_targets_min": 1702 | |
| }, | |
| { | |
| "epoch": 2.376395534290271, | |
| "grad_norm": 0.5858693851449541, | |
| "learning_rate": 3.342196856718074e-05, | |
| "loss": 0.2136, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19342803955078125, | |
| "step": 1490, | |
| "valid_targets_mean": 2573.2, | |
| "valid_targets_min": 684 | |
| }, | |
| { | |
| "epoch": 2.384370015948963, | |
| "grad_norm": 0.664991036812387, | |
| "learning_rate": 3.336289861746602e-05, | |
| "loss": 0.1909, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19867199659347534, | |
| "step": 1495, | |
| "valid_targets_mean": 2357.4, | |
| "valid_targets_min": 907 | |
| }, | |
| { | |
| "epoch": 2.3923444976076556, | |
| "grad_norm": 1.048356992635596, | |
| "learning_rate": 3.330361734528823e-05, | |
| "loss": 0.1955, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20687635242938995, | |
| "step": 1500, | |
| "valid_targets_mean": 3404.9, | |
| "valid_targets_min": 1473 | |
| }, | |
| { | |
| "epoch": 2.4003189792663475, | |
| "grad_norm": 0.6470737570437413, | |
| "learning_rate": 3.324412568812844e-05, | |
| "loss": 0.1723, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14438386261463165, | |
| "step": 1505, | |
| "valid_targets_mean": 1706.2, | |
| "valid_targets_min": 1222 | |
| }, | |
| { | |
| "epoch": 2.40829346092504, | |
| "grad_norm": 0.5670728091943865, | |
| "learning_rate": 3.318442458679477e-05, | |
| "loss": 0.214, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18530724942684174, | |
| "step": 1510, | |
| "valid_targets_mean": 2804.8, | |
| "valid_targets_min": 1128 | |
| }, | |
| { | |
| "epoch": 2.416267942583732, | |
| "grad_norm": 0.5832801162625506, | |
| "learning_rate": 3.312451498540751e-05, | |
| "loss": 0.1739, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17289073765277863, | |
| "step": 1515, | |
| "valid_targets_mean": 2337.8, | |
| "valid_targets_min": 1001 | |
| }, | |
| { | |
| "epoch": 2.4242424242424243, | |
| "grad_norm": 0.5933227626430476, | |
| "learning_rate": 3.306439783138421e-05, | |
| "loss": 0.1829, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16400037705898285, | |
| "step": 1520, | |
| "valid_targets_mean": 2690.1, | |
| "valid_targets_min": 1146 | |
| }, | |
| { | |
| "epoch": 2.4322169059011163, | |
| "grad_norm": 0.5243891317529017, | |
| "learning_rate": 3.3004074075424666e-05, | |
| "loss": 0.1905, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17849090695381165, | |
| "step": 1525, | |
| "valid_targets_mean": 3053.1, | |
| "valid_targets_min": 1151 | |
| }, | |
| { | |
| "epoch": 2.4401913875598087, | |
| "grad_norm": 0.6178786903379436, | |
| "learning_rate": 3.29435446714959e-05, | |
| "loss": 0.2132, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21913208067417145, | |
| "step": 1530, | |
| "valid_targets_mean": 3507.8, | |
| "valid_targets_min": 1474 | |
| }, | |
| { | |
| "epoch": 2.4481658692185007, | |
| "grad_norm": 0.5841079747434857, | |
| "learning_rate": 3.288281057681709e-05, | |
| "loss": 0.183, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17646032571792603, | |
| "step": 1535, | |
| "valid_targets_mean": 2283.2, | |
| "valid_targets_min": 1278 | |
| }, | |
| { | |
| "epoch": 2.456140350877193, | |
| "grad_norm": 0.5168606282768784, | |
| "learning_rate": 3.28218727518444e-05, | |
| "loss": 0.1767, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15758445858955383, | |
| "step": 1540, | |
| "valid_targets_mean": 3063.2, | |
| "valid_targets_min": 1047 | |
| }, | |
| { | |
| "epoch": 2.464114832535885, | |
| "grad_norm": 0.6165157331403959, | |
| "learning_rate": 3.2760732160255835e-05, | |
| "loss": 0.2001, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21482142806053162, | |
| "step": 1545, | |
| "valid_targets_mean": 2569.4, | |
| "valid_targets_min": 1244 | |
| }, | |
| { | |
| "epoch": 2.4720893141945774, | |
| "grad_norm": 0.5845544905089886, | |
| "learning_rate": 3.269938976893595e-05, | |
| "loss": 0.1865, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21223688125610352, | |
| "step": 1550, | |
| "valid_targets_mean": 2661.5, | |
| "valid_targets_min": 1040 | |
| }, | |
| { | |
| "epoch": 2.4800637958532694, | |
| "grad_norm": 0.6284605565912317, | |
| "learning_rate": 3.2637846547960596e-05, | |
| "loss": 0.1994, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17033833265304565, | |
| "step": 1555, | |
| "valid_targets_mean": 2224.4, | |
| "valid_targets_min": 1146 | |
| }, | |
| { | |
| "epoch": 2.488038277511962, | |
| "grad_norm": 0.7490821757292555, | |
| "learning_rate": 3.2576103470581564e-05, | |
| "loss": 0.202, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.31501030921936035, | |
| "step": 1560, | |
| "valid_targets_mean": 3381.4, | |
| "valid_targets_min": 1226 | |
| }, | |
| { | |
| "epoch": 2.496012759170654, | |
| "grad_norm": 0.5814501383804671, | |
| "learning_rate": 3.25141615132112e-05, | |
| "loss": 0.2067, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19716112315654755, | |
| "step": 1565, | |
| "valid_targets_mean": 2829.0, | |
| "valid_targets_min": 1113 | |
| }, | |
| { | |
| "epoch": 2.503987240829346, | |
| "grad_norm": 0.5136642074516043, | |
| "learning_rate": 3.245202165540697e-05, | |
| "loss": 0.1924, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19882862269878387, | |
| "step": 1570, | |
| "valid_targets_mean": 3679.9, | |
| "valid_targets_min": 1278 | |
| }, | |
| { | |
| "epoch": 2.511961722488038, | |
| "grad_norm": 0.5440586455353177, | |
| "learning_rate": 3.238968487985594e-05, | |
| "loss": 0.2069, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19790814816951752, | |
| "step": 1575, | |
| "valid_targets_mean": 2993.3, | |
| "valid_targets_min": 876 | |
| }, | |
| { | |
| "epoch": 2.5199362041467306, | |
| "grad_norm": 0.5929188884939683, | |
| "learning_rate": 3.232715217235927e-05, | |
| "loss": 0.1938, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2090296745300293, | |
| "step": 1580, | |
| "valid_targets_mean": 2928.6, | |
| "valid_targets_min": 1064 | |
| }, | |
| { | |
| "epoch": 2.5279106858054226, | |
| "grad_norm": 0.5697340637761833, | |
| "learning_rate": 3.226442452181662e-05, | |
| "loss": 0.198, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17136168479919434, | |
| "step": 1585, | |
| "valid_targets_mean": 2591.6, | |
| "valid_targets_min": 1261 | |
| }, | |
| { | |
| "epoch": 2.535885167464115, | |
| "grad_norm": 0.540923625766811, | |
| "learning_rate": 3.220150292021049e-05, | |
| "loss": 0.1797, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19687724113464355, | |
| "step": 1590, | |
| "valid_targets_mean": 3221.6, | |
| "valid_targets_min": 1261 | |
| }, | |
| { | |
| "epoch": 2.543859649122807, | |
| "grad_norm": 0.5767857398726057, | |
| "learning_rate": 3.213838836259055e-05, | |
| "loss": 0.1762, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19307035207748413, | |
| "step": 1595, | |
| "valid_targets_mean": 2589.1, | |
| "valid_targets_min": 1046 | |
| }, | |
| { | |
| "epoch": 2.5518341307814993, | |
| "grad_norm": 0.5166304301744661, | |
| "learning_rate": 3.2075081847057886e-05, | |
| "loss": 0.212, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20582881569862366, | |
| "step": 1600, | |
| "valid_targets_mean": 3190.3, | |
| "valid_targets_min": 1186 | |
| }, | |
| { | |
| "epoch": 2.5598086124401913, | |
| "grad_norm": 0.5911901543387457, | |
| "learning_rate": 3.201158437474925e-05, | |
| "loss": 0.2124, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.24600306153297424, | |
| "step": 1605, | |
| "valid_targets_mean": 2878.9, | |
| "valid_targets_min": 961 | |
| }, | |
| { | |
| "epoch": 2.5677830940988837, | |
| "grad_norm": 0.5694511504999716, | |
| "learning_rate": 3.194789694982119e-05, | |
| "loss": 0.1671, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15048804879188538, | |
| "step": 1610, | |
| "valid_targets_mean": 2276.8, | |
| "valid_targets_min": 1519 | |
| }, | |
| { | |
| "epoch": 2.5757575757575757, | |
| "grad_norm": 0.5872429136127384, | |
| "learning_rate": 3.1884020579434216e-05, | |
| "loss": 0.1874, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1900998055934906, | |
| "step": 1615, | |
| "valid_targets_mean": 2539.7, | |
| "valid_targets_min": 1001 | |
| }, | |
| { | |
| "epoch": 2.583732057416268, | |
| "grad_norm": 0.6141904777771472, | |
| "learning_rate": 3.181995627373679e-05, | |
| "loss": 0.187, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16003698110580444, | |
| "step": 1620, | |
| "valid_targets_mean": 1980.6, | |
| "valid_targets_min": 1166 | |
| }, | |
| { | |
| "epoch": 2.59170653907496, | |
| "grad_norm": 0.5646234400091081, | |
| "learning_rate": 3.1755705045849465e-05, | |
| "loss": 0.1829, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15863600373268127, | |
| "step": 1625, | |
| "valid_targets_mean": 2413.1, | |
| "valid_targets_min": 1018 | |
| }, | |
| { | |
| "epoch": 2.5996810207336525, | |
| "grad_norm": 0.5748907862112433, | |
| "learning_rate": 3.1691267911848765e-05, | |
| "loss": 0.1967, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23583777248859406, | |
| "step": 1630, | |
| "valid_targets_mean": 3131.8, | |
| "valid_targets_min": 1268 | |
| }, | |
| { | |
| "epoch": 2.6076555023923444, | |
| "grad_norm": 0.5379326811898751, | |
| "learning_rate": 3.1626645890751167e-05, | |
| "loss": 0.1938, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16717740893363953, | |
| "step": 1635, | |
| "valid_targets_mean": 2830.4, | |
| "valid_targets_min": 1309 | |
| }, | |
| { | |
| "epoch": 2.6156299840510364, | |
| "grad_norm": 0.5334708951980555, | |
| "learning_rate": 3.156184000449697e-05, | |
| "loss": 0.1954, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16978397965431213, | |
| "step": 1640, | |
| "valid_targets_mean": 2473.9, | |
| "valid_targets_min": 1083 | |
| }, | |
| { | |
| "epoch": 2.623604465709729, | |
| "grad_norm": 0.6318420574052661, | |
| "learning_rate": 3.149685127793415e-05, | |
| "loss": 0.1928, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17268671095371246, | |
| "step": 1645, | |
| "valid_targets_mean": 2450.8, | |
| "valid_targets_min": 1123 | |
| }, | |
| { | |
| "epoch": 2.6315789473684212, | |
| "grad_norm": 0.5308821935515915, | |
| "learning_rate": 3.143168073880214e-05, | |
| "loss": 0.2063, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21769240498542786, | |
| "step": 1650, | |
| "valid_targets_mean": 3183.6, | |
| "valid_targets_min": 1088 | |
| }, | |
| { | |
| "epoch": 2.639553429027113, | |
| "grad_norm": 0.6278883319373487, | |
| "learning_rate": 3.1366329417715556e-05, | |
| "loss": 0.1821, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14011511206626892, | |
| "step": 1655, | |
| "valid_targets_mean": 1544.2, | |
| "valid_targets_min": 986 | |
| }, | |
| { | |
| "epoch": 2.647527910685805, | |
| "grad_norm": 0.5913767520819432, | |
| "learning_rate": 3.1300798348147954e-05, | |
| "loss": 0.1871, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17621494829654694, | |
| "step": 1660, | |
| "valid_targets_mean": 2409.2, | |
| "valid_targets_min": 1282 | |
| }, | |
| { | |
| "epoch": 2.6555023923444976, | |
| "grad_norm": 0.5708954991838666, | |
| "learning_rate": 3.123508856641542e-05, | |
| "loss": 0.1691, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20209786295890808, | |
| "step": 1665, | |
| "valid_targets_mean": 3007.8, | |
| "valid_targets_min": 1544 | |
| }, | |
| { | |
| "epoch": 2.66347687400319, | |
| "grad_norm": 0.6239083317789432, | |
| "learning_rate": 3.116920111166025e-05, | |
| "loss": 0.183, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14001812040805817, | |
| "step": 1670, | |
| "valid_targets_mean": 2042.8, | |
| "valid_targets_min": 1135 | |
| }, | |
| { | |
| "epoch": 2.671451355661882, | |
| "grad_norm": 0.5555429521600216, | |
| "learning_rate": 3.1103137025834456e-05, | |
| "loss": 0.1862, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1607971489429474, | |
| "step": 1675, | |
| "valid_targets_mean": 2579.4, | |
| "valid_targets_min": 1481 | |
| }, | |
| { | |
| "epoch": 2.679425837320574, | |
| "grad_norm": 0.5911984658901864, | |
| "learning_rate": 3.103689735368333e-05, | |
| "loss": 0.179, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1880512535572052, | |
| "step": 1680, | |
| "valid_targets_mean": 2316.4, | |
| "valid_targets_min": 1078 | |
| }, | |
| { | |
| "epoch": 2.6874003189792663, | |
| "grad_norm": 0.5568950096417259, | |
| "learning_rate": 3.097048314272889e-05, | |
| "loss": 0.1724, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14909867942333221, | |
| "step": 1685, | |
| "valid_targets_mean": 2620.3, | |
| "valid_targets_min": 1035 | |
| }, | |
| { | |
| "epoch": 2.6953748006379588, | |
| "grad_norm": 0.5605104059179533, | |
| "learning_rate": 3.090389544325335e-05, | |
| "loss": 0.1887, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19391119480133057, | |
| "step": 1690, | |
| "valid_targets_mean": 3403.4, | |
| "valid_targets_min": 906 | |
| }, | |
| { | |
| "epoch": 2.7033492822966507, | |
| "grad_norm": 0.5940720122887997, | |
| "learning_rate": 3.08371353082825e-05, | |
| "loss": 0.179, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21021555364131927, | |
| "step": 1695, | |
| "valid_targets_mean": 2653.8, | |
| "valid_targets_min": 1347 | |
| }, | |
| { | |
| "epoch": 2.7113237639553427, | |
| "grad_norm": 0.5279976704245198, | |
| "learning_rate": 3.0770203793568994e-05, | |
| "loss": 0.2179, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20088006556034088, | |
| "step": 1700, | |
| "valid_targets_mean": 3224.9, | |
| "valid_targets_min": 1552 | |
| }, | |
| { | |
| "epoch": 2.719298245614035, | |
| "grad_norm": 0.5728810440642814, | |
| "learning_rate": 3.0703101957575765e-05, | |
| "loss": 0.199, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16884160041809082, | |
| "step": 1705, | |
| "valid_targets_mean": 2506.7, | |
| "valid_targets_min": 1038 | |
| }, | |
| { | |
| "epoch": 2.7272727272727275, | |
| "grad_norm": 0.5630970898895722, | |
| "learning_rate": 3.0635830861459204e-05, | |
| "loss": 0.2041, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.201943039894104, | |
| "step": 1710, | |
| "valid_targets_mean": 3317.2, | |
| "valid_targets_min": 1394 | |
| }, | |
| { | |
| "epoch": 2.7352472089314195, | |
| "grad_norm": 0.5401867778634583, | |
| "learning_rate": 3.056839156905239e-05, | |
| "loss": 0.1777, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17215926945209503, | |
| "step": 1715, | |
| "valid_targets_mean": 2514.6, | |
| "valid_targets_min": 550 | |
| }, | |
| { | |
| "epoch": 2.7432216905901115, | |
| "grad_norm": 0.7680039863293809, | |
| "learning_rate": 3.0500785146848303e-05, | |
| "loss": 0.1897, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1482524424791336, | |
| "step": 1720, | |
| "valid_targets_mean": 2329.0, | |
| "valid_targets_min": 1025 | |
| }, | |
| { | |
| "epoch": 2.751196172248804, | |
| "grad_norm": 0.5728974492985291, | |
| "learning_rate": 3.04330126639829e-05, | |
| "loss": 0.2008, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15376368165016174, | |
| "step": 1725, | |
| "valid_targets_mean": 2507.7, | |
| "valid_targets_min": 1303 | |
| }, | |
| { | |
| "epoch": 2.7591706539074963, | |
| "grad_norm": 0.5664221776878716, | |
| "learning_rate": 3.0365075192218278e-05, | |
| "loss": 0.1836, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1676475703716278, | |
| "step": 1730, | |
| "valid_targets_mean": 2541.6, | |
| "valid_targets_min": 1031 | |
| }, | |
| { | |
| "epoch": 2.7671451355661882, | |
| "grad_norm": 0.5049114348515064, | |
| "learning_rate": 3.0296973805925663e-05, | |
| "loss": 0.1852, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18172882497310638, | |
| "step": 1735, | |
| "valid_targets_mean": 3568.7, | |
| "valid_targets_min": 1451 | |
| }, | |
| { | |
| "epoch": 2.77511961722488, | |
| "grad_norm": 0.5614918409668362, | |
| "learning_rate": 3.022870958206845e-05, | |
| "loss": 0.1933, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18978315591812134, | |
| "step": 1740, | |
| "valid_targets_mean": 2648.4, | |
| "valid_targets_min": 1236 | |
| }, | |
| { | |
| "epoch": 2.7830940988835726, | |
| "grad_norm": 0.5542031035808245, | |
| "learning_rate": 3.0160283600185168e-05, | |
| "loss": 0.1731, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1841156929731369, | |
| "step": 1745, | |
| "valid_targets_mean": 3271.1, | |
| "valid_targets_min": 1346 | |
| }, | |
| { | |
| "epoch": 2.7910685805422646, | |
| "grad_norm": 0.568674574848384, | |
| "learning_rate": 3.0091696942372412e-05, | |
| "loss": 0.2149, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22050027549266815, | |
| "step": 1750, | |
| "valid_targets_mean": 2670.2, | |
| "valid_targets_min": 1306 | |
| }, | |
| { | |
| "epoch": 2.799043062200957, | |
| "grad_norm": 0.5643138963885681, | |
| "learning_rate": 3.002295069326772e-05, | |
| "loss": 0.1826, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18906863033771515, | |
| "step": 1755, | |
| "valid_targets_mean": 2565.9, | |
| "valid_targets_min": 1377 | |
| }, | |
| { | |
| "epoch": 2.807017543859649, | |
| "grad_norm": 0.6646069591582359, | |
| "learning_rate": 2.9954045940032423e-05, | |
| "loss": 0.1874, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23465055227279663, | |
| "step": 1760, | |
| "valid_targets_mean": 3034.7, | |
| "valid_targets_min": 888 | |
| }, | |
| { | |
| "epoch": 2.8149920255183414, | |
| "grad_norm": 0.5196774748614063, | |
| "learning_rate": 2.988498377233446e-05, | |
| "loss": 0.1892, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2014995664358139, | |
| "step": 1765, | |
| "valid_targets_mean": 3231.3, | |
| "valid_targets_min": 1098 | |
| }, | |
| { | |
| "epoch": 2.8229665071770333, | |
| "grad_norm": 0.5502942878993454, | |
| "learning_rate": 2.981576528233114e-05, | |
| "loss": 0.1926, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2188304364681244, | |
| "step": 1770, | |
| "valid_targets_mean": 3371.0, | |
| "valid_targets_min": 1076 | |
| }, | |
| { | |
| "epoch": 2.8309409888357258, | |
| "grad_norm": 0.5425762676905854, | |
| "learning_rate": 2.9746391564651867e-05, | |
| "loss": 0.1713, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14157424867153168, | |
| "step": 1775, | |
| "valid_targets_mean": 2150.8, | |
| "valid_targets_min": 1266 | |
| }, | |
| { | |
| "epoch": 2.8389154704944177, | |
| "grad_norm": 0.6014818751525163, | |
| "learning_rate": 2.9676863716380845e-05, | |
| "loss": 0.201, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1941225528717041, | |
| "step": 1780, | |
| "valid_targets_mean": 4106.5, | |
| "valid_targets_min": 1109 | |
| }, | |
| { | |
| "epoch": 2.84688995215311, | |
| "grad_norm": 0.5538284681110416, | |
| "learning_rate": 2.9607182837039697e-05, | |
| "loss": 0.1841, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16050055623054504, | |
| "step": 1785, | |
| "valid_targets_mean": 2549.5, | |
| "valid_targets_min": 1034 | |
| }, | |
| { | |
| "epoch": 2.854864433811802, | |
| "grad_norm": 0.6724791781520713, | |
| "learning_rate": 2.9537350028570126e-05, | |
| "loss": 0.2148, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23556634783744812, | |
| "step": 1790, | |
| "valid_targets_mean": 2610.4, | |
| "valid_targets_min": 880 | |
| }, | |
| { | |
| "epoch": 2.8628389154704945, | |
| "grad_norm": 0.5282103567404306, | |
| "learning_rate": 2.946736639531643e-05, | |
| "loss": 0.2098, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19032198190689087, | |
| "step": 1795, | |
| "valid_targets_mean": 3423.3, | |
| "valid_targets_min": 840 | |
| }, | |
| { | |
| "epoch": 2.8708133971291865, | |
| "grad_norm": 0.5343948181525061, | |
| "learning_rate": 2.9397233044008092e-05, | |
| "loss": 0.1815, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16319958865642548, | |
| "step": 1800, | |
| "valid_targets_mean": 2711.1, | |
| "valid_targets_min": 1324 | |
| }, | |
| { | |
| "epoch": 2.878787878787879, | |
| "grad_norm": 0.6069494403984073, | |
| "learning_rate": 2.9326951083742243e-05, | |
| "loss": 0.1797, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15907859802246094, | |
| "step": 1805, | |
| "valid_targets_mean": 2519.6, | |
| "valid_targets_min": 1252 | |
| }, | |
| { | |
| "epoch": 2.886762360446571, | |
| "grad_norm": 0.5831177287355487, | |
| "learning_rate": 2.925652162596613e-05, | |
| "loss": 0.1946, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22586268186569214, | |
| "step": 1810, | |
| "valid_targets_mean": 3239.4, | |
| "valid_targets_min": 1232 | |
| }, | |
| { | |
| "epoch": 2.8947368421052633, | |
| "grad_norm": 0.5605548506235492, | |
| "learning_rate": 2.9185945784459558e-05, | |
| "loss": 0.1812, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17734205722808838, | |
| "step": 1815, | |
| "valid_targets_mean": 2670.1, | |
| "valid_targets_min": 1058 | |
| }, | |
| { | |
| "epoch": 2.9027113237639552, | |
| "grad_norm": 0.6194121532589498, | |
| "learning_rate": 2.9115224675317252e-05, | |
| "loss": 0.1719, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17597773671150208, | |
| "step": 1820, | |
| "valid_targets_mean": 1934.1, | |
| "valid_targets_min": 994 | |
| }, | |
| { | |
| "epoch": 2.9106858054226477, | |
| "grad_norm": 0.5784067931656794, | |
| "learning_rate": 2.9044359416931206e-05, | |
| "loss": 0.1911, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2120431363582611, | |
| "step": 1825, | |
| "valid_targets_mean": 3082.4, | |
| "valid_targets_min": 894 | |
| }, | |
| { | |
| "epoch": 2.9186602870813396, | |
| "grad_norm": 0.5518771187801405, | |
| "learning_rate": 2.897335112997302e-05, | |
| "loss": 0.2055, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2005378007888794, | |
| "step": 1830, | |
| "valid_targets_mean": 2880.1, | |
| "valid_targets_min": 896 | |
| }, | |
| { | |
| "epoch": 2.926634768740032, | |
| "grad_norm": 0.48456097909556345, | |
| "learning_rate": 2.8902200937376173e-05, | |
| "loss": 0.1823, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15114593505859375, | |
| "step": 1835, | |
| "valid_targets_mean": 3304.1, | |
| "valid_targets_min": 601 | |
| }, | |
| { | |
| "epoch": 2.934609250398724, | |
| "grad_norm": 0.568638742445266, | |
| "learning_rate": 2.8830909964318242e-05, | |
| "loss": 0.2012, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.197839617729187, | |
| "step": 1840, | |
| "valid_targets_mean": 3082.9, | |
| "valid_targets_min": 1170 | |
| }, | |
| { | |
| "epoch": 2.9425837320574164, | |
| "grad_norm": 0.6009143002900112, | |
| "learning_rate": 2.875947933820312e-05, | |
| "loss": 0.1691, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21282166242599487, | |
| "step": 1845, | |
| "valid_targets_mean": 2744.5, | |
| "valid_targets_min": 868 | |
| }, | |
| { | |
| "epoch": 2.9505582137161084, | |
| "grad_norm": 0.5835820573311307, | |
| "learning_rate": 2.868791018864321e-05, | |
| "loss": 0.1913, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23180079460144043, | |
| "step": 1850, | |
| "valid_targets_mean": 3296.6, | |
| "valid_targets_min": 1201 | |
| }, | |
| { | |
| "epoch": 2.958532695374801, | |
| "grad_norm": 0.5403850798020928, | |
| "learning_rate": 2.861620364744151e-05, | |
| "loss": 0.188, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2239145040512085, | |
| "step": 1855, | |
| "valid_targets_mean": 3232.8, | |
| "valid_targets_min": 984 | |
| }, | |
| { | |
| "epoch": 2.9665071770334928, | |
| "grad_norm": 0.5471992005694861, | |
| "learning_rate": 2.854436084857379e-05, | |
| "loss": 0.2042, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20272240042686462, | |
| "step": 1860, | |
| "valid_targets_mean": 2604.9, | |
| "valid_targets_min": 1635 | |
| }, | |
| { | |
| "epoch": 2.974481658692185, | |
| "grad_norm": 0.5866640715776678, | |
| "learning_rate": 2.847238292817057e-05, | |
| "loss": 0.1913, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20167194306850433, | |
| "step": 1865, | |
| "valid_targets_mean": 2490.1, | |
| "valid_targets_min": 1066 | |
| }, | |
| { | |
| "epoch": 2.982456140350877, | |
| "grad_norm": 0.5904315155737803, | |
| "learning_rate": 2.8400271024499212e-05, | |
| "loss": 0.1692, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18105053901672363, | |
| "step": 1870, | |
| "valid_targets_mean": 2653.9, | |
| "valid_targets_min": 1072 | |
| }, | |
| { | |
| "epoch": 2.990430622009569, | |
| "grad_norm": 0.5667528402230967, | |
| "learning_rate": 2.832802627794593e-05, | |
| "loss": 0.1999, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22661757469177246, | |
| "step": 1875, | |
| "valid_targets_mean": 3706.2, | |
| "valid_targets_min": 1121 | |
| }, | |
| { | |
| "epoch": 2.9984051036682615, | |
| "grad_norm": 0.6009440796314518, | |
| "learning_rate": 2.8255649830997704e-05, | |
| "loss": 0.1878, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20299099385738373, | |
| "step": 1880, | |
| "valid_targets_mean": 3680.2, | |
| "valid_targets_min": 961 | |
| }, | |
| { | |
| "epoch": 3.006379585326954, | |
| "grad_norm": 0.6099874287063509, | |
| "learning_rate": 2.8183142828224238e-05, | |
| "loss": 0.1726, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1845843642950058, | |
| "step": 1885, | |
| "valid_targets_mean": 2441.4, | |
| "valid_targets_min": 1443 | |
| }, | |
| { | |
| "epoch": 3.014354066985646, | |
| "grad_norm": 0.6051159039153519, | |
| "learning_rate": 2.811050641625988e-05, | |
| "loss": 0.1734, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18158197402954102, | |
| "step": 1890, | |
| "valid_targets_mean": 2382.5, | |
| "valid_targets_min": 1009 | |
| }, | |
| { | |
| "epoch": 3.0223285486443383, | |
| "grad_norm": 0.5721617590448416, | |
| "learning_rate": 2.8037741743785458e-05, | |
| "loss": 0.1751, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16156092286109924, | |
| "step": 1895, | |
| "valid_targets_mean": 2683.9, | |
| "valid_targets_min": 1074 | |
| }, | |
| { | |
| "epoch": 3.0303030303030303, | |
| "grad_norm": 0.5865329434183887, | |
| "learning_rate": 2.796484996151013e-05, | |
| "loss": 0.1907, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11619686335325241, | |
| "step": 1900, | |
| "valid_targets_mean": 2044.0, | |
| "valid_targets_min": 1046 | |
| }, | |
| { | |
| "epoch": 3.0382775119617227, | |
| "grad_norm": 0.5737857689212434, | |
| "learning_rate": 2.7891832222153188e-05, | |
| "loss": 0.1658, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1338677853345871, | |
| "step": 1905, | |
| "valid_targets_mean": 2470.9, | |
| "valid_targets_min": 994 | |
| }, | |
| { | |
| "epoch": 3.0462519936204147, | |
| "grad_norm": 0.5108539805469473, | |
| "learning_rate": 2.7818689680425807e-05, | |
| "loss": 0.1816, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17531079053878784, | |
| "step": 1910, | |
| "valid_targets_mean": 3671.6, | |
| "valid_targets_min": 1240 | |
| }, | |
| { | |
| "epoch": 3.054226475279107, | |
| "grad_norm": 0.5556796605107793, | |
| "learning_rate": 2.7745423493012827e-05, | |
| "loss": 0.1715, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18171335756778717, | |
| "step": 1915, | |
| "valid_targets_mean": 2726.2, | |
| "valid_targets_min": 961 | |
| }, | |
| { | |
| "epoch": 3.062200956937799, | |
| "grad_norm": 0.6178579891937267, | |
| "learning_rate": 2.767203481855441e-05, | |
| "loss": 0.1763, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18336257338523865, | |
| "step": 1920, | |
| "valid_targets_mean": 2685.5, | |
| "valid_targets_min": 1271 | |
| }, | |
| { | |
| "epoch": 3.0701754385964914, | |
| "grad_norm": 0.5854283357456007, | |
| "learning_rate": 2.7598524817627764e-05, | |
| "loss": 0.1931, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19227474927902222, | |
| "step": 1925, | |
| "valid_targets_mean": 2520.4, | |
| "valid_targets_min": 1045 | |
| }, | |
| { | |
| "epoch": 3.0781499202551834, | |
| "grad_norm": 0.49625404123998523, | |
| "learning_rate": 2.7524894652728754e-05, | |
| "loss": 0.1822, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19637757539749146, | |
| "step": 1930, | |
| "valid_targets_mean": 4134.2, | |
| "valid_targets_min": 1222 | |
| }, | |
| { | |
| "epoch": 3.0861244019138754, | |
| "grad_norm": 0.5676068839717044, | |
| "learning_rate": 2.7451145488253544e-05, | |
| "loss": 0.1962, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22619007527828217, | |
| "step": 1935, | |
| "valid_targets_mean": 3551.0, | |
| "valid_targets_min": 1108 | |
| }, | |
| { | |
| "epoch": 3.094098883572568, | |
| "grad_norm": 0.7692746963023798, | |
| "learning_rate": 2.7377278490480157e-05, | |
| "loss": 0.1784, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17253446578979492, | |
| "step": 1940, | |
| "valid_targets_mean": 3062.4, | |
| "valid_targets_min": 1077 | |
| }, | |
| { | |
| "epoch": 3.1020733652312598, | |
| "grad_norm": 0.5526740615386009, | |
| "learning_rate": 2.730329482755006e-05, | |
| "loss": 0.1623, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20209446549415588, | |
| "step": 1945, | |
| "valid_targets_mean": 2935.2, | |
| "valid_targets_min": 1001 | |
| }, | |
| { | |
| "epoch": 3.110047846889952, | |
| "grad_norm": 0.5510587896491016, | |
| "learning_rate": 2.7229195669449667e-05, | |
| "loss": 0.1863, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1594051718711853, | |
| "step": 1950, | |
| "valid_targets_mean": 3520.5, | |
| "valid_targets_min": 920 | |
| }, | |
| { | |
| "epoch": 3.118022328548644, | |
| "grad_norm": 0.6250742491422023, | |
| "learning_rate": 2.7154982187991855e-05, | |
| "loss": 0.2154, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2715832591056824, | |
| "step": 1955, | |
| "valid_targets_mean": 3909.7, | |
| "valid_targets_min": 1282 | |
| }, | |
| { | |
| "epoch": 3.1259968102073366, | |
| "grad_norm": 0.5920653083228358, | |
| "learning_rate": 2.7080655556797406e-05, | |
| "loss": 0.1904, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.26988518238067627, | |
| "step": 1960, | |
| "valid_targets_mean": 3452.6, | |
| "valid_targets_min": 712 | |
| }, | |
| { | |
| "epoch": 3.1339712918660285, | |
| "grad_norm": 0.5388774310856131, | |
| "learning_rate": 2.700621695127649e-05, | |
| "loss": 0.1769, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15851232409477234, | |
| "step": 1965, | |
| "valid_targets_mean": 2872.0, | |
| "valid_targets_min": 1072 | |
| }, | |
| { | |
| "epoch": 3.141945773524721, | |
| "grad_norm": 0.6691557242877121, | |
| "learning_rate": 2.693166754861003e-05, | |
| "loss": 0.1857, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2240079641342163, | |
| "step": 1970, | |
| "valid_targets_mean": 2786.1, | |
| "valid_targets_min": 1217 | |
| }, | |
| { | |
| "epoch": 3.149920255183413, | |
| "grad_norm": 0.6366719433898753, | |
| "learning_rate": 2.685700852773113e-05, | |
| "loss": 0.1999, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19620132446289062, | |
| "step": 1975, | |
| "valid_targets_mean": 2583.2, | |
| "valid_targets_min": 1304 | |
| }, | |
| { | |
| "epoch": 3.1578947368421053, | |
| "grad_norm": 0.5928192806363577, | |
| "learning_rate": 2.6782241069306395e-05, | |
| "loss": 0.1782, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.186061292886734, | |
| "step": 1980, | |
| "valid_targets_mean": 2680.7, | |
| "valid_targets_min": 1416 | |
| }, | |
| { | |
| "epoch": 3.1658692185007973, | |
| "grad_norm": 0.5855724054324057, | |
| "learning_rate": 2.6707366355717268e-05, | |
| "loss": 0.1733, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21362203359603882, | |
| "step": 1985, | |
| "valid_targets_mean": 2905.8, | |
| "valid_targets_min": 965 | |
| }, | |
| { | |
| "epoch": 3.1738437001594897, | |
| "grad_norm": 0.6347489114308262, | |
| "learning_rate": 2.663238557104136e-05, | |
| "loss": 0.1715, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1517581045627594, | |
| "step": 1990, | |
| "valid_targets_mean": 2070.0, | |
| "valid_targets_min": 1324 | |
| }, | |
| { | |
| "epoch": 3.1818181818181817, | |
| "grad_norm": 0.5635841038270237, | |
| "learning_rate": 2.655729990103368e-05, | |
| "loss": 0.1966, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1979377269744873, | |
| "step": 1995, | |
| "valid_targets_mean": 2958.0, | |
| "valid_targets_min": 1064 | |
| }, | |
| { | |
| "epoch": 3.189792663476874, | |
| "grad_norm": 0.5945458950462633, | |
| "learning_rate": 2.648211053310792e-05, | |
| "loss": 0.1708, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15361088514328003, | |
| "step": 2000, | |
| "valid_targets_mean": 2395.7, | |
| "valid_targets_min": 1268 | |
| }, | |
| { | |
| "epoch": 3.197767145135566, | |
| "grad_norm": 0.6299192799682408, | |
| "learning_rate": 2.6406818656317654e-05, | |
| "loss": 0.1635, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15608029067516327, | |
| "step": 2005, | |
| "valid_targets_mean": 2129.3, | |
| "valid_targets_min": 692 | |
| }, | |
| { | |
| "epoch": 3.2057416267942584, | |
| "grad_norm": 0.6201929767056361, | |
| "learning_rate": 2.6331425461337557e-05, | |
| "loss": 0.1746, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19049637019634247, | |
| "step": 2010, | |
| "valid_targets_mean": 2675.4, | |
| "valid_targets_min": 1237 | |
| }, | |
| { | |
| "epoch": 3.2137161084529504, | |
| "grad_norm": 0.5993976587951823, | |
| "learning_rate": 2.6255932140444546e-05, | |
| "loss": 0.1995, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20035934448242188, | |
| "step": 2015, | |
| "valid_targets_mean": 2800.6, | |
| "valid_targets_min": 1342 | |
| }, | |
| { | |
| "epoch": 3.221690590111643, | |
| "grad_norm": 0.5460305244294847, | |
| "learning_rate": 2.618033988749895e-05, | |
| "loss": 0.1697, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17547008395195007, | |
| "step": 2020, | |
| "valid_targets_mean": 3045.4, | |
| "valid_targets_min": 1199 | |
| }, | |
| { | |
| "epoch": 3.229665071770335, | |
| "grad_norm": 0.5515986645818807, | |
| "learning_rate": 2.6104649897925622e-05, | |
| "loss": 0.1679, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17502939701080322, | |
| "step": 2025, | |
| "valid_targets_mean": 2976.9, | |
| "valid_targets_min": 1040 | |
| }, | |
| { | |
| "epoch": 3.237639553429027, | |
| "grad_norm": 0.5334926575974487, | |
| "learning_rate": 2.602886336869503e-05, | |
| "loss": 0.164, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13687369227409363, | |
| "step": 2030, | |
| "valid_targets_mean": 2390.6, | |
| "valid_targets_min": 743 | |
| }, | |
| { | |
| "epoch": 3.245614035087719, | |
| "grad_norm": 0.5518442305418664, | |
| "learning_rate": 2.595298149830433e-05, | |
| "loss": 0.1714, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16078054904937744, | |
| "step": 2035, | |
| "valid_targets_mean": 2902.9, | |
| "valid_targets_min": 703 | |
| }, | |
| { | |
| "epoch": 3.2535885167464116, | |
| "grad_norm": 0.5690058283856204, | |
| "learning_rate": 2.5877005486758424e-05, | |
| "loss": 0.1928, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18337497115135193, | |
| "step": 2040, | |
| "valid_targets_mean": 2966.4, | |
| "valid_targets_min": 840 | |
| }, | |
| { | |
| "epoch": 3.2615629984051036, | |
| "grad_norm": 0.497799727420445, | |
| "learning_rate": 2.5800936535550954e-05, | |
| "loss": 0.2149, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1807919144630432, | |
| "step": 2045, | |
| "valid_targets_mean": 3695.0, | |
| "valid_targets_min": 1274 | |
| }, | |
| { | |
| "epoch": 3.269537480063796, | |
| "grad_norm": 0.544150458920545, | |
| "learning_rate": 2.5724775847645345e-05, | |
| "loss": 0.2194, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.283038467168808, | |
| "step": 2050, | |
| "valid_targets_mean": 3945.1, | |
| "valid_targets_min": 1421 | |
| }, | |
| { | |
| "epoch": 3.277511961722488, | |
| "grad_norm": 0.4913399971466565, | |
| "learning_rate": 2.5648524627455738e-05, | |
| "loss": 0.1976, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19701933860778809, | |
| "step": 2055, | |
| "valid_targets_mean": 4155.9, | |
| "valid_targets_min": 694 | |
| }, | |
| { | |
| "epoch": 3.2854864433811803, | |
| "grad_norm": 0.5691728034505411, | |
| "learning_rate": 2.557218408082798e-05, | |
| "loss": 0.182, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18620505928993225, | |
| "step": 2060, | |
| "valid_targets_mean": 2885.2, | |
| "valid_targets_min": 1029 | |
| }, | |
| { | |
| "epoch": 3.2934609250398723, | |
| "grad_norm": 0.5111010875857084, | |
| "learning_rate": 2.5495755415020504e-05, | |
| "loss": 0.1878, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1496405005455017, | |
| "step": 2065, | |
| "valid_targets_mean": 2963.8, | |
| "valid_targets_min": 1239 | |
| }, | |
| { | |
| "epoch": 3.3014354066985647, | |
| "grad_norm": 0.5407424289252192, | |
| "learning_rate": 2.5419239838685325e-05, | |
| "loss": 0.1686, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15042206645011902, | |
| "step": 2070, | |
| "valid_targets_mean": 3061.2, | |
| "valid_targets_min": 1108 | |
| }, | |
| { | |
| "epoch": 3.3094098883572567, | |
| "grad_norm": 0.5973954022258182, | |
| "learning_rate": 2.5342638561848817e-05, | |
| "loss": 0.1869, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1537877321243286, | |
| "step": 2075, | |
| "valid_targets_mean": 2506.9, | |
| "valid_targets_min": 1150 | |
| }, | |
| { | |
| "epoch": 3.317384370015949, | |
| "grad_norm": 0.5830150859025254, | |
| "learning_rate": 2.526595279589265e-05, | |
| "loss": 0.1769, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1255737543106079, | |
| "step": 2080, | |
| "valid_targets_mean": 2168.4, | |
| "valid_targets_min": 1331 | |
| }, | |
| { | |
| "epoch": 3.325358851674641, | |
| "grad_norm": 0.5136748644322553, | |
| "learning_rate": 2.5189183753534634e-05, | |
| "loss": 0.1728, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17841193079948425, | |
| "step": 2085, | |
| "valid_targets_mean": 3524.8, | |
| "valid_targets_min": 1194 | |
| }, | |
| { | |
| "epoch": 3.3333333333333335, | |
| "grad_norm": 0.5444148283630629, | |
| "learning_rate": 2.5112332648809495e-05, | |
| "loss": 0.182, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19738224148750305, | |
| "step": 2090, | |
| "valid_targets_mean": 3265.0, | |
| "valid_targets_min": 1305 | |
| }, | |
| { | |
| "epoch": 3.3413078149920254, | |
| "grad_norm": 0.5736105889034541, | |
| "learning_rate": 2.503540069704973e-05, | |
| "loss": 0.1875, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18508929014205933, | |
| "step": 2095, | |
| "valid_targets_mean": 3098.3, | |
| "valid_targets_min": 835 | |
| }, | |
| { | |
| "epoch": 3.349282296650718, | |
| "grad_norm": 0.5444972670187521, | |
| "learning_rate": 2.4958389114866326e-05, | |
| "loss": 0.2084, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23350661993026733, | |
| "step": 2100, | |
| "valid_targets_mean": 3487.3, | |
| "valid_targets_min": 878 | |
| }, | |
| { | |
| "epoch": 3.35725677830941, | |
| "grad_norm": 0.607507746990593, | |
| "learning_rate": 2.488129912012958e-05, | |
| "loss": 0.1902, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1646861582994461, | |
| "step": 2105, | |
| "valid_targets_mean": 2289.1, | |
| "valid_targets_min": 1059 | |
| }, | |
| { | |
| "epoch": 3.3652312599681022, | |
| "grad_norm": 0.5590542729013029, | |
| "learning_rate": 2.4804131931949823e-05, | |
| "loss": 0.1718, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1932382434606552, | |
| "step": 2110, | |
| "valid_targets_mean": 3424.1, | |
| "valid_targets_min": 993 | |
| }, | |
| { | |
| "epoch": 3.373205741626794, | |
| "grad_norm": 0.5792226727709087, | |
| "learning_rate": 2.4726888770658103e-05, | |
| "loss": 0.1645, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1489265263080597, | |
| "step": 2115, | |
| "valid_targets_mean": 2261.2, | |
| "valid_targets_min": 1268 | |
| }, | |
| { | |
| "epoch": 3.3811802232854866, | |
| "grad_norm": 0.6496431060482095, | |
| "learning_rate": 2.4649570857786928e-05, | |
| "loss": 0.1789, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15027418732643127, | |
| "step": 2120, | |
| "valid_targets_mean": 2156.9, | |
| "valid_targets_min": 876 | |
| }, | |
| { | |
| "epoch": 3.3891547049441786, | |
| "grad_norm": 0.5528609427449673, | |
| "learning_rate": 2.4572179416050953e-05, | |
| "loss": 0.1887, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16369466483592987, | |
| "step": 2125, | |
| "valid_targets_mean": 2808.6, | |
| "valid_targets_min": 1234 | |
| }, | |
| { | |
| "epoch": 3.397129186602871, | |
| "grad_norm": 0.5312391691731903, | |
| "learning_rate": 2.449471566932761e-05, | |
| "loss": 0.1894, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17131400108337402, | |
| "step": 2130, | |
| "valid_targets_mean": 3416.1, | |
| "valid_targets_min": 1042 | |
| }, | |
| { | |
| "epoch": 3.405103668261563, | |
| "grad_norm": 0.6228851428878182, | |
| "learning_rate": 2.4417180842637764e-05, | |
| "loss": 0.176, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23682790994644165, | |
| "step": 2135, | |
| "valid_targets_mean": 3038.9, | |
| "valid_targets_min": 1159 | |
| }, | |
| { | |
| "epoch": 3.4130781499202554, | |
| "grad_norm": 0.5882304013111794, | |
| "learning_rate": 2.4339576162126362e-05, | |
| "loss": 0.1763, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15148214995861053, | |
| "step": 2140, | |
| "valid_targets_mean": 2432.0, | |
| "valid_targets_min": 248 | |
| }, | |
| { | |
| "epoch": 3.4210526315789473, | |
| "grad_norm": 0.5783651613849465, | |
| "learning_rate": 2.4261902855043027e-05, | |
| "loss": 0.1703, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17573174834251404, | |
| "step": 2145, | |
| "valid_targets_mean": 2605.1, | |
| "valid_targets_min": 1267 | |
| }, | |
| { | |
| "epoch": 3.4290271132376393, | |
| "grad_norm": 0.6002324738524472, | |
| "learning_rate": 2.418416214972265e-05, | |
| "loss": 0.1628, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1507042646408081, | |
| "step": 2150, | |
| "valid_targets_mean": 1904.8, | |
| "valid_targets_min": 1031 | |
| }, | |
| { | |
| "epoch": 3.4370015948963317, | |
| "grad_norm": 0.49264136447640344, | |
| "learning_rate": 2.4106355275565937e-05, | |
| "loss": 0.2018, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1505666971206665, | |
| "step": 2155, | |
| "valid_targets_mean": 2797.3, | |
| "valid_targets_min": 1417 | |
| }, | |
| { | |
| "epoch": 3.444976076555024, | |
| "grad_norm": 0.5621913458885389, | |
| "learning_rate": 2.4028483463020053e-05, | |
| "loss": 0.1813, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16950352489948273, | |
| "step": 2160, | |
| "valid_targets_mean": 2679.9, | |
| "valid_targets_min": 1238 | |
| }, | |
| { | |
| "epoch": 3.452950558213716, | |
| "grad_norm": 0.5401249898056637, | |
| "learning_rate": 2.3950547943559056e-05, | |
| "loss": 0.1908, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22693544626235962, | |
| "step": 2165, | |
| "valid_targets_mean": 3421.9, | |
| "valid_targets_min": 1372 | |
| }, | |
| { | |
| "epoch": 3.460925039872408, | |
| "grad_norm": 0.6699395433332309, | |
| "learning_rate": 2.3872549949664493e-05, | |
| "loss": 0.185, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16658160090446472, | |
| "step": 2170, | |
| "valid_targets_mean": 2192.9, | |
| "valid_targets_min": 999 | |
| }, | |
| { | |
| "epoch": 3.4688995215311005, | |
| "grad_norm": 0.6254529715179462, | |
| "learning_rate": 2.3794490714805897e-05, | |
| "loss": 0.192, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.186039999127388, | |
| "step": 2175, | |
| "valid_targets_mean": 2131.5, | |
| "valid_targets_min": 894 | |
| }, | |
| { | |
| "epoch": 3.476874003189793, | |
| "grad_norm": 0.5627992056782284, | |
| "learning_rate": 2.3716371473421242e-05, | |
| "loss": 0.1666, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15978986024856567, | |
| "step": 2180, | |
| "valid_targets_mean": 2397.3, | |
| "valid_targets_min": 1089 | |
| }, | |
| { | |
| "epoch": 3.484848484848485, | |
| "grad_norm": 0.5869604883871188, | |
| "learning_rate": 2.36381934608975e-05, | |
| "loss": 0.1974, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18033649027347565, | |
| "step": 2185, | |
| "valid_targets_mean": 2868.6, | |
| "valid_targets_min": 1227 | |
| }, | |
| { | |
| "epoch": 3.492822966507177, | |
| "grad_norm": 0.5069371015301846, | |
| "learning_rate": 2.3559957913551014e-05, | |
| "loss": 0.1632, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19203317165374756, | |
| "step": 2190, | |
| "valid_targets_mean": 4175.7, | |
| "valid_targets_min": 1122 | |
| }, | |
| { | |
| "epoch": 3.5007974481658692, | |
| "grad_norm": 0.6745474518471709, | |
| "learning_rate": 2.3481666068607994e-05, | |
| "loss": 0.1905, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18692749738693237, | |
| "step": 2195, | |
| "valid_targets_mean": 2161.5, | |
| "valid_targets_min": 1216 | |
| }, | |
| { | |
| "epoch": 3.5087719298245617, | |
| "grad_norm": 0.5614763734838379, | |
| "learning_rate": 2.3403319164184986e-05, | |
| "loss": 0.1931, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1870293766260147, | |
| "step": 2200, | |
| "valid_targets_mean": 2776.0, | |
| "valid_targets_min": 1081 | |
| }, | |
| { | |
| "epoch": 3.5167464114832536, | |
| "grad_norm": 0.5970202439939211, | |
| "learning_rate": 2.3324918439269206e-05, | |
| "loss": 0.1943, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16551634669303894, | |
| "step": 2205, | |
| "valid_targets_mean": 2269.2, | |
| "valid_targets_min": 632 | |
| }, | |
| { | |
| "epoch": 3.5247208931419456, | |
| "grad_norm": 0.621103572086824, | |
| "learning_rate": 2.3246465133699024e-05, | |
| "loss": 0.173, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16670405864715576, | |
| "step": 2210, | |
| "valid_targets_mean": 2442.2, | |
| "valid_targets_min": 1039 | |
| }, | |
| { | |
| "epoch": 3.532695374800638, | |
| "grad_norm": 0.49934228070645226, | |
| "learning_rate": 2.31679604881443e-05, | |
| "loss": 0.1742, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18905407190322876, | |
| "step": 2215, | |
| "valid_targets_mean": 3872.6, | |
| "valid_targets_min": 1423 | |
| }, | |
| { | |
| "epoch": 3.5406698564593304, | |
| "grad_norm": 0.6269225467887087, | |
| "learning_rate": 2.3089405744086823e-05, | |
| "loss": 0.1595, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16947981715202332, | |
| "step": 2220, | |
| "valid_targets_mean": 2441.1, | |
| "valid_targets_min": 925 | |
| }, | |
| { | |
| "epoch": 3.5486443381180224, | |
| "grad_norm": 0.5286942755540239, | |
| "learning_rate": 2.3010802143800626e-05, | |
| "loss": 0.166, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1366422474384308, | |
| "step": 2225, | |
| "valid_targets_mean": 2366.1, | |
| "valid_targets_min": 1002 | |
| }, | |
| { | |
| "epoch": 3.5566188197767143, | |
| "grad_norm": 0.5109251022213953, | |
| "learning_rate": 2.2932150930332363e-05, | |
| "loss": 0.1746, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14900361001491547, | |
| "step": 2230, | |
| "valid_targets_mean": 3294.4, | |
| "valid_targets_min": 1424 | |
| }, | |
| { | |
| "epoch": 3.5645933014354068, | |
| "grad_norm": 0.6397212335830404, | |
| "learning_rate": 2.2853453347481635e-05, | |
| "loss": 0.1729, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1566600203514099, | |
| "step": 2235, | |
| "valid_targets_mean": 2303.5, | |
| "valid_targets_min": 601 | |
| }, | |
| { | |
| "epoch": 3.5725677830940987, | |
| "grad_norm": 0.5801376421441098, | |
| "learning_rate": 2.277471063978137e-05, | |
| "loss": 0.1673, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1630595624446869, | |
| "step": 2240, | |
| "valid_targets_mean": 2375.6, | |
| "valid_targets_min": 1396 | |
| }, | |
| { | |
| "epoch": 3.580542264752791, | |
| "grad_norm": 0.5790198614924473, | |
| "learning_rate": 2.2695924052478065e-05, | |
| "loss": 0.1879, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17185233533382416, | |
| "step": 2245, | |
| "valid_targets_mean": 2518.6, | |
| "valid_targets_min": 1198 | |
| }, | |
| { | |
| "epoch": 3.588516746411483, | |
| "grad_norm": 0.5243705768816054, | |
| "learning_rate": 2.2617094831512167e-05, | |
| "loss": 0.196, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20980407297611237, | |
| "step": 2250, | |
| "valid_targets_mean": 3963.8, | |
| "valid_targets_min": 1345 | |
| }, | |
| { | |
| "epoch": 3.5964912280701755, | |
| "grad_norm": 0.5504953256074006, | |
| "learning_rate": 2.253822422349831e-05, | |
| "loss": 0.2063, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18440350890159607, | |
| "step": 2255, | |
| "valid_targets_mean": 2997.2, | |
| "valid_targets_min": 1185 | |
| }, | |
| { | |
| "epoch": 3.6044657097288675, | |
| "grad_norm": 0.5728779102295528, | |
| "learning_rate": 2.2459313475705645e-05, | |
| "loss": 0.1805, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16484129428863525, | |
| "step": 2260, | |
| "valid_targets_mean": 2457.5, | |
| "valid_targets_min": 1256 | |
| }, | |
| { | |
| "epoch": 3.61244019138756, | |
| "grad_norm": 0.6105030719650987, | |
| "learning_rate": 2.23803638360381e-05, | |
| "loss": 0.1913, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16316813230514526, | |
| "step": 2265, | |
| "valid_targets_mean": 2532.6, | |
| "valid_targets_min": 862 | |
| }, | |
| { | |
| "epoch": 3.620414673046252, | |
| "grad_norm": 0.5665339579502443, | |
| "learning_rate": 2.2301376553014625e-05, | |
| "loss": 0.1764, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16501827538013458, | |
| "step": 2270, | |
| "valid_targets_mean": 2997.6, | |
| "valid_targets_min": 1141 | |
| }, | |
| { | |
| "epoch": 3.6283891547049443, | |
| "grad_norm": 0.6138831066285112, | |
| "learning_rate": 2.2222352875749493e-05, | |
| "loss": 0.1817, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19114506244659424, | |
| "step": 2275, | |
| "valid_targets_mean": 3209.3, | |
| "valid_targets_min": 1469 | |
| }, | |
| { | |
| "epoch": 3.6363636363636362, | |
| "grad_norm": 0.4615867272467092, | |
| "learning_rate": 2.214329405393249e-05, | |
| "loss": 0.1888, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1981082260608673, | |
| "step": 2280, | |
| "valid_targets_mean": 4500.8, | |
| "valid_targets_min": 1297 | |
| }, | |
| { | |
| "epoch": 3.6443381180223287, | |
| "grad_norm": 0.5442457635323399, | |
| "learning_rate": 2.2064201337809212e-05, | |
| "loss": 0.1545, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1776713728904724, | |
| "step": 2285, | |
| "valid_targets_mean": 3355.9, | |
| "valid_targets_min": 1223 | |
| }, | |
| { | |
| "epoch": 3.6523125996810206, | |
| "grad_norm": 0.5407377473015359, | |
| "learning_rate": 2.1985075978161236e-05, | |
| "loss": 0.1828, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15134361386299133, | |
| "step": 2290, | |
| "valid_targets_mean": 2769.6, | |
| "valid_targets_min": 1416 | |
| }, | |
| { | |
| "epoch": 3.660287081339713, | |
| "grad_norm": 0.5905714881664845, | |
| "learning_rate": 2.1905919226286385e-05, | |
| "loss": 0.1681, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13224008679389954, | |
| "step": 2295, | |
| "valid_targets_mean": 2251.6, | |
| "valid_targets_min": 1285 | |
| }, | |
| { | |
| "epoch": 3.668261562998405, | |
| "grad_norm": 0.6215279831334835, | |
| "learning_rate": 2.1826732333978924e-05, | |
| "loss": 0.1731, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22618988156318665, | |
| "step": 2300, | |
| "valid_targets_mean": 2626.7, | |
| "valid_targets_min": 1169 | |
| }, | |
| { | |
| "epoch": 3.6762360446570974, | |
| "grad_norm": 0.5696379458737713, | |
| "learning_rate": 2.1747516553509755e-05, | |
| "loss": 0.179, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13785511255264282, | |
| "step": 2305, | |
| "valid_targets_mean": 2088.3, | |
| "valid_targets_min": 1069 | |
| }, | |
| { | |
| "epoch": 3.6842105263157894, | |
| "grad_norm": 0.5709846807235273, | |
| "learning_rate": 2.166827313760662e-05, | |
| "loss": 0.19, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1877063363790512, | |
| "step": 2310, | |
| "valid_targets_mean": 2509.9, | |
| "valid_targets_min": 1049 | |
| }, | |
| { | |
| "epoch": 3.692185007974482, | |
| "grad_norm": 0.546669144933103, | |
| "learning_rate": 2.1589003339434292e-05, | |
| "loss": 0.181, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21933197975158691, | |
| "step": 2315, | |
| "valid_targets_mean": 3100.3, | |
| "valid_targets_min": 1135 | |
| }, | |
| { | |
| "epoch": 3.7001594896331738, | |
| "grad_norm": 0.5776670004886075, | |
| "learning_rate": 2.150970841257476e-05, | |
| "loss": 0.1907, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21792183816432953, | |
| "step": 2320, | |
| "valid_targets_mean": 3190.8, | |
| "valid_targets_min": 1066 | |
| }, | |
| { | |
| "epoch": 3.708133971291866, | |
| "grad_norm": 0.5740982492996695, | |
| "learning_rate": 2.1430389611007393e-05, | |
| "loss": 0.1814, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16459020972251892, | |
| "step": 2325, | |
| "valid_targets_mean": 2925.2, | |
| "valid_targets_min": 1140 | |
| }, | |
| { | |
| "epoch": 3.716108452950558, | |
| "grad_norm": 0.5443262921569451, | |
| "learning_rate": 2.135104818908913e-05, | |
| "loss": 0.1633, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16006751358509064, | |
| "step": 2330, | |
| "valid_targets_mean": 2719.9, | |
| "valid_targets_min": 894 | |
| }, | |
| { | |
| "epoch": 3.7240829346092506, | |
| "grad_norm": 0.5979662168113598, | |
| "learning_rate": 2.1271685401534617e-05, | |
| "loss": 0.1753, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18777906894683838, | |
| "step": 2335, | |
| "valid_targets_mean": 2444.4, | |
| "valid_targets_min": 886 | |
| }, | |
| { | |
| "epoch": 3.7320574162679425, | |
| "grad_norm": 0.5280536723772452, | |
| "learning_rate": 2.11923025033964e-05, | |
| "loss": 0.1554, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12486009299755096, | |
| "step": 2340, | |
| "valid_targets_mean": 2508.4, | |
| "valid_targets_min": 1081 | |
| }, | |
| { | |
| "epoch": 3.740031897926635, | |
| "grad_norm": 0.5014224094405413, | |
| "learning_rate": 2.111290075004503e-05, | |
| "loss": 0.1758, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2104087471961975, | |
| "step": 2345, | |
| "valid_targets_mean": 4018.2, | |
| "valid_targets_min": 1134 | |
| }, | |
| { | |
| "epoch": 3.748006379585327, | |
| "grad_norm": 0.5329277393739597, | |
| "learning_rate": 2.103348139714925e-05, | |
| "loss": 0.172, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15234363079071045, | |
| "step": 2350, | |
| "valid_targets_mean": 2975.2, | |
| "valid_targets_min": 1209 | |
| }, | |
| { | |
| "epoch": 3.7559808612440193, | |
| "grad_norm": 0.602906573401119, | |
| "learning_rate": 2.0954045700656126e-05, | |
| "loss": 0.1873, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23616360127925873, | |
| "step": 2355, | |
| "valid_targets_mean": 2888.1, | |
| "valid_targets_min": 723 | |
| }, | |
| { | |
| "epoch": 3.7639553429027113, | |
| "grad_norm": 0.593349469128777, | |
| "learning_rate": 2.0874594916771174e-05, | |
| "loss": 0.1933, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17083615064620972, | |
| "step": 2360, | |
| "valid_targets_mean": 2501.6, | |
| "valid_targets_min": 1053 | |
| }, | |
| { | |
| "epoch": 3.7719298245614032, | |
| "grad_norm": 0.6659895307842664, | |
| "learning_rate": 2.079513030193852e-05, | |
| "loss": 0.1944, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18279269337654114, | |
| "step": 2365, | |
| "valid_targets_mean": 3181.6, | |
| "valid_targets_min": 1554 | |
| }, | |
| { | |
| "epoch": 3.7799043062200957, | |
| "grad_norm": 0.5481485498020948, | |
| "learning_rate": 2.071565311282099e-05, | |
| "loss": 0.193, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15074336528778076, | |
| "step": 2370, | |
| "valid_targets_mean": 2559.0, | |
| "valid_targets_min": 1116 | |
| }, | |
| { | |
| "epoch": 3.787878787878788, | |
| "grad_norm": 0.5064359267566692, | |
| "learning_rate": 2.063616460628029e-05, | |
| "loss": 0.1822, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13388487696647644, | |
| "step": 2375, | |
| "valid_targets_mean": 2798.5, | |
| "valid_targets_min": 1149 | |
| }, | |
| { | |
| "epoch": 3.79585326953748, | |
| "grad_norm": 0.6015447662377474, | |
| "learning_rate": 2.0556666039357084e-05, | |
| "loss": 0.1584, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1580061912536621, | |
| "step": 2380, | |
| "valid_targets_mean": 2257.2, | |
| "valid_targets_min": 1265 | |
| }, | |
| { | |
| "epoch": 3.803827751196172, | |
| "grad_norm": 0.6425996831220174, | |
| "learning_rate": 2.047715866925113e-05, | |
| "loss": 0.1682, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17762739956378937, | |
| "step": 2385, | |
| "valid_targets_mean": 2541.7, | |
| "valid_targets_min": 990 | |
| }, | |
| { | |
| "epoch": 3.8118022328548644, | |
| "grad_norm": 0.6128275854039842, | |
| "learning_rate": 2.0397643753301403e-05, | |
| "loss": 0.1688, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1614629626274109, | |
| "step": 2390, | |
| "valid_targets_mean": 2946.4, | |
| "valid_targets_min": 1119 | |
| }, | |
| { | |
| "epoch": 3.819776714513557, | |
| "grad_norm": 0.5505144745464942, | |
| "learning_rate": 2.0318122548966225e-05, | |
| "loss": 0.1845, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21430611610412598, | |
| "step": 2395, | |
| "valid_targets_mean": 3036.6, | |
| "valid_targets_min": 1255 | |
| }, | |
| { | |
| "epoch": 3.827751196172249, | |
| "grad_norm": 0.4407896880987488, | |
| "learning_rate": 2.0238596313803337e-05, | |
| "loss": 0.18, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17456912994384766, | |
| "step": 2400, | |
| "valid_targets_mean": 4507.5, | |
| "valid_targets_min": 1037 | |
| }, | |
| { | |
| "epoch": 3.8357256778309408, | |
| "grad_norm": 0.533906121005407, | |
| "learning_rate": 2.015906630545005e-05, | |
| "loss": 0.1949, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19968262314796448, | |
| "step": 2405, | |
| "valid_targets_mean": 3266.3, | |
| "valid_targets_min": 965 | |
| }, | |
| { | |
| "epoch": 3.843700159489633, | |
| "grad_norm": 0.6010100638921084, | |
| "learning_rate": 2.0079533781603352e-05, | |
| "loss": 0.1916, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15822017192840576, | |
| "step": 2410, | |
| "valid_targets_mean": 2278.6, | |
| "valid_targets_min": 1337 | |
| }, | |
| { | |
| "epoch": 3.8516746411483256, | |
| "grad_norm": 0.6656969404639518, | |
| "learning_rate": 2e-05, | |
| "loss": 0.1752, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15312731266021729, | |
| "step": 2415, | |
| "valid_targets_mean": 1937.1, | |
| "valid_targets_min": 946 | |
| }, | |
| { | |
| "epoch": 3.8596491228070176, | |
| "grad_norm": 0.5908822557428244, | |
| "learning_rate": 1.9920466218396655e-05, | |
| "loss": 0.1915, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1849689483642578, | |
| "step": 2420, | |
| "valid_targets_mean": 2551.8, | |
| "valid_targets_min": 984 | |
| }, | |
| { | |
| "epoch": 3.8676236044657095, | |
| "grad_norm": 0.5868172639371974, | |
| "learning_rate": 1.9840933694549956e-05, | |
| "loss": 0.1736, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16618412733078003, | |
| "step": 2425, | |
| "valid_targets_mean": 2423.7, | |
| "valid_targets_min": 978 | |
| }, | |
| { | |
| "epoch": 3.875598086124402, | |
| "grad_norm": 0.5929987209389399, | |
| "learning_rate": 1.976140368619667e-05, | |
| "loss": 0.2043, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2104598581790924, | |
| "step": 2430, | |
| "valid_targets_mean": 2879.4, | |
| "valid_targets_min": 1403 | |
| }, | |
| { | |
| "epoch": 3.8835725677830943, | |
| "grad_norm": 0.5696736725853189, | |
| "learning_rate": 1.9681877451033778e-05, | |
| "loss": 0.2107, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1889794021844864, | |
| "step": 2435, | |
| "valid_targets_mean": 2941.5, | |
| "valid_targets_min": 1374 | |
| }, | |
| { | |
| "epoch": 3.8915470494417863, | |
| "grad_norm": 0.8300816587532014, | |
| "learning_rate": 1.9602356246698597e-05, | |
| "loss": 0.1736, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22295412421226501, | |
| "step": 2440, | |
| "valid_targets_mean": 2925.1, | |
| "valid_targets_min": 1288 | |
| }, | |
| { | |
| "epoch": 3.8995215311004783, | |
| "grad_norm": 0.5343275886193569, | |
| "learning_rate": 1.9522841330748877e-05, | |
| "loss": 0.1828, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1584158092737198, | |
| "step": 2445, | |
| "valid_targets_mean": 2872.1, | |
| "valid_targets_min": 1526 | |
| }, | |
| { | |
| "epoch": 3.9074960127591707, | |
| "grad_norm": 0.5571151028017529, | |
| "learning_rate": 1.944333396064292e-05, | |
| "loss": 0.1929, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2206161618232727, | |
| "step": 2450, | |
| "valid_targets_mean": 3363.2, | |
| "valid_targets_min": 999 | |
| }, | |
| { | |
| "epoch": 3.915470494417863, | |
| "grad_norm": 0.557234488585679, | |
| "learning_rate": 1.936383539371971e-05, | |
| "loss": 0.1916, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22577141225337982, | |
| "step": 2455, | |
| "valid_targets_mean": 3063.4, | |
| "valid_targets_min": 1112 | |
| }, | |
| { | |
| "epoch": 3.923444976076555, | |
| "grad_norm": 0.6518598913086615, | |
| "learning_rate": 1.9284346887179016e-05, | |
| "loss": 0.1791, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18474088609218597, | |
| "step": 2460, | |
| "valid_targets_mean": 2250.4, | |
| "valid_targets_min": 1063 | |
| }, | |
| { | |
| "epoch": 3.931419457735247, | |
| "grad_norm": 0.591784233490105, | |
| "learning_rate": 1.9204869698061493e-05, | |
| "loss": 0.1761, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18672320246696472, | |
| "step": 2465, | |
| "valid_targets_mean": 2717.3, | |
| "valid_targets_min": 1039 | |
| }, | |
| { | |
| "epoch": 3.9393939393939394, | |
| "grad_norm": 0.6029208831187829, | |
| "learning_rate": 1.9125405083228833e-05, | |
| "loss": 0.2183, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21028904616832733, | |
| "step": 2470, | |
| "valid_targets_mean": 2852.6, | |
| "valid_targets_min": 1309 | |
| }, | |
| { | |
| "epoch": 3.9473684210526314, | |
| "grad_norm": 0.5681207981391441, | |
| "learning_rate": 1.9045954299343884e-05, | |
| "loss": 0.1561, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1799524873495102, | |
| "step": 2475, | |
| "valid_targets_mean": 3248.3, | |
| "valid_targets_min": 1519 | |
| }, | |
| { | |
| "epoch": 3.955342902711324, | |
| "grad_norm": 0.5531686772161876, | |
| "learning_rate": 1.896651860285076e-05, | |
| "loss": 0.1977, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17599454522132874, | |
| "step": 2480, | |
| "valid_targets_mean": 2960.8, | |
| "valid_targets_min": 996 | |
| }, | |
| { | |
| "epoch": 3.963317384370016, | |
| "grad_norm": 0.567746336035401, | |
| "learning_rate": 1.8887099249954976e-05, | |
| "loss": 0.2041, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19742515683174133, | |
| "step": 2485, | |
| "valid_targets_mean": 2644.5, | |
| "valid_targets_min": 1062 | |
| }, | |
| { | |
| "epoch": 3.971291866028708, | |
| "grad_norm": 0.5926141200227866, | |
| "learning_rate": 1.8807697496603604e-05, | |
| "loss": 0.1575, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16229791939258575, | |
| "step": 2490, | |
| "valid_targets_mean": 2446.4, | |
| "valid_targets_min": 1201 | |
| }, | |
| { | |
| "epoch": 3.9792663476874, | |
| "grad_norm": 0.5559012262231059, | |
| "learning_rate": 1.8728314598465386e-05, | |
| "loss": 0.1884, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1845509111881256, | |
| "step": 2495, | |
| "valid_targets_mean": 3171.1, | |
| "valid_targets_min": 1375 | |
| }, | |
| { | |
| "epoch": 3.9872408293460926, | |
| "grad_norm": 0.6154642770548927, | |
| "learning_rate": 1.8648951810910878e-05, | |
| "loss": 0.1851, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15992483496665955, | |
| "step": 2500, | |
| "valid_targets_mean": 2345.1, | |
| "valid_targets_min": 1146 | |
| }, | |
| { | |
| "epoch": 3.9952153110047846, | |
| "grad_norm": 0.5570345470466502, | |
| "learning_rate": 1.856961038899261e-05, | |
| "loss": 0.1961, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21707212924957275, | |
| "step": 2505, | |
| "valid_targets_mean": 3223.9, | |
| "valid_targets_min": 1235 | |
| }, | |
| { | |
| "epoch": 4.003189792663477, | |
| "grad_norm": 0.4820862025485545, | |
| "learning_rate": 1.849029158742525e-05, | |
| "loss": 0.1583, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13848713040351868, | |
| "step": 2510, | |
| "valid_targets_mean": 3489.1, | |
| "valid_targets_min": 1517 | |
| }, | |
| { | |
| "epoch": 4.011164274322169, | |
| "grad_norm": 0.5449115879256036, | |
| "learning_rate": 1.8410996660565714e-05, | |
| "loss": 0.1578, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17959289252758026, | |
| "step": 2515, | |
| "valid_targets_mean": 3114.4, | |
| "valid_targets_min": 1154 | |
| }, | |
| { | |
| "epoch": 4.019138755980861, | |
| "grad_norm": 0.5821484435562528, | |
| "learning_rate": 1.8331726862393385e-05, | |
| "loss": 0.1503, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15955208241939545, | |
| "step": 2520, | |
| "valid_targets_mean": 3422.6, | |
| "valid_targets_min": 1333 | |
| }, | |
| { | |
| "epoch": 4.027113237639553, | |
| "grad_norm": 0.5333494672871051, | |
| "learning_rate": 1.8252483446490245e-05, | |
| "loss": 0.1661, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16043981909751892, | |
| "step": 2525, | |
| "valid_targets_mean": 3289.8, | |
| "valid_targets_min": 1258 | |
| }, | |
| { | |
| "epoch": 4.035087719298246, | |
| "grad_norm": 0.681384185389914, | |
| "learning_rate": 1.8173267666021075e-05, | |
| "loss": 0.1773, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15066920220851898, | |
| "step": 2530, | |
| "valid_targets_mean": 1848.1, | |
| "valid_targets_min": 1058 | |
| }, | |
| { | |
| "epoch": 4.043062200956938, | |
| "grad_norm": 0.5926951676146437, | |
| "learning_rate": 1.8094080773713615e-05, | |
| "loss": 0.1686, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1673404574394226, | |
| "step": 2535, | |
| "valid_targets_mean": 2794.1, | |
| "valid_targets_min": 1031 | |
| }, | |
| { | |
| "epoch": 4.05103668261563, | |
| "grad_norm": 0.6585195947195234, | |
| "learning_rate": 1.8014924021838774e-05, | |
| "loss": 0.1585, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15531276166439056, | |
| "step": 2540, | |
| "valid_targets_mean": 2173.6, | |
| "valid_targets_min": 1112 | |
| }, | |
| { | |
| "epoch": 4.059011164274322, | |
| "grad_norm": 0.6431155766625963, | |
| "learning_rate": 1.7935798662190798e-05, | |
| "loss": 0.177, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20385132730007172, | |
| "step": 2545, | |
| "valid_targets_mean": 3026.9, | |
| "valid_targets_min": 880 | |
| }, | |
| { | |
| "epoch": 4.0669856459330145, | |
| "grad_norm": 0.6055270197304871, | |
| "learning_rate": 1.785670594606752e-05, | |
| "loss": 0.1612, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16407035291194916, | |
| "step": 2550, | |
| "valid_targets_mean": 2962.2, | |
| "valid_targets_min": 1299 | |
| }, | |
| { | |
| "epoch": 4.074960127591707, | |
| "grad_norm": 0.527292972057754, | |
| "learning_rate": 1.777764712425052e-05, | |
| "loss": 0.1645, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1590946912765503, | |
| "step": 2555, | |
| "valid_targets_mean": 3571.1, | |
| "valid_targets_min": 424 | |
| }, | |
| { | |
| "epoch": 4.082934609250398, | |
| "grad_norm": 0.5778748659086131, | |
| "learning_rate": 1.7698623446985378e-05, | |
| "loss": 0.1482, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1918186992406845, | |
| "step": 2560, | |
| "valid_targets_mean": 3398.8, | |
| "valid_targets_min": 1252 | |
| }, | |
| { | |
| "epoch": 4.090909090909091, | |
| "grad_norm": 0.5230022966524009, | |
| "learning_rate": 1.7619636163961907e-05, | |
| "loss": 0.1461, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16492609679698944, | |
| "step": 2565, | |
| "valid_targets_mean": 3574.9, | |
| "valid_targets_min": 1142 | |
| }, | |
| { | |
| "epoch": 4.098883572567783, | |
| "grad_norm": 0.600059132410499, | |
| "learning_rate": 1.754068652429436e-05, | |
| "loss": 0.1761, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22227483987808228, | |
| "step": 2570, | |
| "valid_targets_mean": 3160.7, | |
| "valid_targets_min": 1115 | |
| }, | |
| { | |
| "epoch": 4.106858054226476, | |
| "grad_norm": 0.5506153806974453, | |
| "learning_rate": 1.74617757765017e-05, | |
| "loss": 0.1534, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15465757250785828, | |
| "step": 2575, | |
| "valid_targets_mean": 3507.8, | |
| "valid_targets_min": 1169 | |
| }, | |
| { | |
| "epoch": 4.114832535885167, | |
| "grad_norm": 0.5872195929772528, | |
| "learning_rate": 1.738290516848784e-05, | |
| "loss": 0.1656, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1792100965976715, | |
| "step": 2580, | |
| "valid_targets_mean": 3182.6, | |
| "valid_targets_min": 1057 | |
| }, | |
| { | |
| "epoch": 4.12280701754386, | |
| "grad_norm": 0.5568326980647963, | |
| "learning_rate": 1.730407594752194e-05, | |
| "loss": 0.1571, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13268157839775085, | |
| "step": 2585, | |
| "valid_targets_mean": 2716.8, | |
| "valid_targets_min": 1001 | |
| }, | |
| { | |
| "epoch": 4.130781499202552, | |
| "grad_norm": 0.5337621603486079, | |
| "learning_rate": 1.722528936021864e-05, | |
| "loss": 0.1655, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13221563398838043, | |
| "step": 2590, | |
| "valid_targets_mean": 2776.6, | |
| "valid_targets_min": 1069 | |
| }, | |
| { | |
| "epoch": 4.138755980861244, | |
| "grad_norm": 0.594154146656907, | |
| "learning_rate": 1.714654665251837e-05, | |
| "loss": 0.1696, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2037222981452942, | |
| "step": 2595, | |
| "valid_targets_mean": 3069.6, | |
| "valid_targets_min": 1025 | |
| }, | |
| { | |
| "epoch": 4.146730462519936, | |
| "grad_norm": 0.5991382655130797, | |
| "learning_rate": 1.7067849069667644e-05, | |
| "loss": 0.173, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1336292177438736, | |
| "step": 2600, | |
| "valid_targets_mean": 2691.4, | |
| "valid_targets_min": 1362 | |
| }, | |
| { | |
| "epoch": 4.154704944178628, | |
| "grad_norm": 0.6333054733885867, | |
| "learning_rate": 1.6989197856199377e-05, | |
| "loss": 0.1518, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12979856133460999, | |
| "step": 2605, | |
| "valid_targets_mean": 2420.9, | |
| "valid_targets_min": 1066 | |
| }, | |
| { | |
| "epoch": 4.162679425837321, | |
| "grad_norm": 0.5693331370434527, | |
| "learning_rate": 1.6910594255913177e-05, | |
| "loss": 0.157, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1348324865102768, | |
| "step": 2610, | |
| "valid_targets_mean": 2557.1, | |
| "valid_targets_min": 1357 | |
| }, | |
| { | |
| "epoch": 4.170653907496013, | |
| "grad_norm": 0.6158456534730181, | |
| "learning_rate": 1.6832039511855702e-05, | |
| "loss": 0.1607, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17839965224266052, | |
| "step": 2615, | |
| "valid_targets_mean": 2557.9, | |
| "valid_targets_min": 1123 | |
| }, | |
| { | |
| "epoch": 4.178628389154705, | |
| "grad_norm": 0.6569117165395834, | |
| "learning_rate": 1.6753534866300983e-05, | |
| "loss": 0.1783, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12349331378936768, | |
| "step": 2620, | |
| "valid_targets_mean": 2012.1, | |
| "valid_targets_min": 1135 | |
| }, | |
| { | |
| "epoch": 4.186602870813397, | |
| "grad_norm": 0.5649615489956892, | |
| "learning_rate": 1.6675081560730804e-05, | |
| "loss": 0.1766, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15348169207572937, | |
| "step": 2625, | |
| "valid_targets_mean": 3120.9, | |
| "valid_targets_min": 993 | |
| }, | |
| { | |
| "epoch": 4.1945773524720895, | |
| "grad_norm": 0.6662040972505084, | |
| "learning_rate": 1.6596680835815024e-05, | |
| "loss": 0.1659, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1765637993812561, | |
| "step": 2630, | |
| "valid_targets_mean": 2654.2, | |
| "valid_targets_min": 1123 | |
| }, | |
| { | |
| "epoch": 4.202551834130782, | |
| "grad_norm": 0.6225824529089881, | |
| "learning_rate": 1.651833393139201e-05, | |
| "loss": 0.161, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14078602194786072, | |
| "step": 2635, | |
| "valid_targets_mean": 3025.2, | |
| "valid_targets_min": 627 | |
| }, | |
| { | |
| "epoch": 4.2105263157894735, | |
| "grad_norm": 0.6675069947682166, | |
| "learning_rate": 1.6440042086449e-05, | |
| "loss": 0.1606, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14914357662200928, | |
| "step": 2640, | |
| "valid_targets_mean": 2228.6, | |
| "valid_targets_min": 1135 | |
| }, | |
| { | |
| "epoch": 4.218500797448166, | |
| "grad_norm": 0.5649692767485978, | |
| "learning_rate": 1.6361806539102508e-05, | |
| "loss": 0.1762, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22645841538906097, | |
| "step": 2645, | |
| "valid_targets_mean": 3634.6, | |
| "valid_targets_min": 1352 | |
| }, | |
| { | |
| "epoch": 4.226475279106858, | |
| "grad_norm": 0.5594120744991358, | |
| "learning_rate": 1.628362852657876e-05, | |
| "loss": 0.1617, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.133091002702713, | |
| "step": 2650, | |
| "valid_targets_mean": 2490.9, | |
| "valid_targets_min": 1209 | |
| }, | |
| { | |
| "epoch": 4.23444976076555, | |
| "grad_norm": 0.6130688435516675, | |
| "learning_rate": 1.620550928519411e-05, | |
| "loss": 0.1712, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17605170607566833, | |
| "step": 2655, | |
| "valid_targets_mean": 3088.2, | |
| "valid_targets_min": 1289 | |
| }, | |
| { | |
| "epoch": 4.242424242424242, | |
| "grad_norm": 0.5599311671798608, | |
| "learning_rate": 1.612745005033551e-05, | |
| "loss": 0.1748, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20437359809875488, | |
| "step": 2660, | |
| "valid_targets_mean": 3273.9, | |
| "valid_targets_min": 1369 | |
| }, | |
| { | |
| "epoch": 4.250398724082935, | |
| "grad_norm": 0.524405728232288, | |
| "learning_rate": 1.6049452056440948e-05, | |
| "loss": 0.1468, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1535736620426178, | |
| "step": 2665, | |
| "valid_targets_mean": 3528.8, | |
| "valid_targets_min": 1250 | |
| }, | |
| { | |
| "epoch": 4.258373205741627, | |
| "grad_norm": 0.9543428316028962, | |
| "learning_rate": 1.597151653697995e-05, | |
| "loss": 0.1987, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12014397233724594, | |
| "step": 2670, | |
| "valid_targets_mean": 1957.2, | |
| "valid_targets_min": 1201 | |
| }, | |
| { | |
| "epoch": 4.266347687400319, | |
| "grad_norm": 0.6675333452530134, | |
| "learning_rate": 1.5893644724434066e-05, | |
| "loss": 0.1601, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17045184969902039, | |
| "step": 2675, | |
| "valid_targets_mean": 2527.7, | |
| "valid_targets_min": 1129 | |
| }, | |
| { | |
| "epoch": 4.274322169059011, | |
| "grad_norm": 0.5530104226801451, | |
| "learning_rate": 1.581583785027736e-05, | |
| "loss": 0.1801, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15394771099090576, | |
| "step": 2680, | |
| "valid_targets_mean": 3375.9, | |
| "valid_targets_min": 1485 | |
| }, | |
| { | |
| "epoch": 4.282296650717703, | |
| "grad_norm": 0.6198052315537499, | |
| "learning_rate": 1.5738097144956976e-05, | |
| "loss": 0.174, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18288853764533997, | |
| "step": 2685, | |
| "valid_targets_mean": 2741.9, | |
| "valid_targets_min": 1101 | |
| }, | |
| { | |
| "epoch": 4.290271132376396, | |
| "grad_norm": 0.5705456528708921, | |
| "learning_rate": 1.566042383787364e-05, | |
| "loss": 0.1635, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17937612533569336, | |
| "step": 2690, | |
| "valid_targets_mean": 3222.2, | |
| "valid_targets_min": 1166 | |
| }, | |
| { | |
| "epoch": 4.298245614035087, | |
| "grad_norm": 0.6678666574110236, | |
| "learning_rate": 1.558281915736224e-05, | |
| "loss": 0.1595, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17287465929985046, | |
| "step": 2695, | |
| "valid_targets_mean": 2454.9, | |
| "valid_targets_min": 1066 | |
| }, | |
| { | |
| "epoch": 4.30622009569378, | |
| "grad_norm": 0.58460956144151, | |
| "learning_rate": 1.550528433067239e-05, | |
| "loss": 0.1507, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15777446329593658, | |
| "step": 2700, | |
| "valid_targets_mean": 2841.6, | |
| "valid_targets_min": 1330 | |
| }, | |
| { | |
| "epoch": 4.314194577352472, | |
| "grad_norm": 0.6720009602673392, | |
| "learning_rate": 1.5427820583949054e-05, | |
| "loss": 0.1498, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17180287837982178, | |
| "step": 2705, | |
| "valid_targets_mean": 2462.7, | |
| "valid_targets_min": 1155 | |
| }, | |
| { | |
| "epoch": 4.3221690590111645, | |
| "grad_norm": 0.6831596441857424, | |
| "learning_rate": 1.5350429142213075e-05, | |
| "loss": 0.1611, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1586819291114807, | |
| "step": 2710, | |
| "valid_targets_mean": 1872.5, | |
| "valid_targets_min": 1235 | |
| }, | |
| { | |
| "epoch": 4.330143540669856, | |
| "grad_norm": 0.5344175912658572, | |
| "learning_rate": 1.5273111229341907e-05, | |
| "loss": 0.1575, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18215686082839966, | |
| "step": 2715, | |
| "valid_targets_mean": 3179.9, | |
| "valid_targets_min": 1334 | |
| }, | |
| { | |
| "epoch": 4.3381180223285485, | |
| "grad_norm": 0.6216985426268544, | |
| "learning_rate": 1.5195868068050185e-05, | |
| "loss": 0.1585, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1297648698091507, | |
| "step": 2720, | |
| "valid_targets_mean": 2275.3, | |
| "valid_targets_min": 1262 | |
| }, | |
| { | |
| "epoch": 4.346092503987241, | |
| "grad_norm": 0.6239029857349583, | |
| "learning_rate": 1.5118700879870426e-05, | |
| "loss": 0.1673, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13424813747406006, | |
| "step": 2725, | |
| "valid_targets_mean": 2379.4, | |
| "valid_targets_min": 1273 | |
| }, | |
| { | |
| "epoch": 4.354066985645933, | |
| "grad_norm": 0.5723646011985536, | |
| "learning_rate": 1.5041610885133681e-05, | |
| "loss": 0.1635, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16782692074775696, | |
| "step": 2730, | |
| "valid_targets_mean": 3187.6, | |
| "valid_targets_min": 1287 | |
| }, | |
| { | |
| "epoch": 4.362041467304625, | |
| "grad_norm": 0.5813610738573333, | |
| "learning_rate": 1.496459930295028e-05, | |
| "loss": 0.1532, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15391327440738678, | |
| "step": 2735, | |
| "valid_targets_mean": 2778.6, | |
| "valid_targets_min": 1408 | |
| }, | |
| { | |
| "epoch": 4.370015948963317, | |
| "grad_norm": 0.658300475272945, | |
| "learning_rate": 1.4887667351190508e-05, | |
| "loss": 0.165, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15785977244377136, | |
| "step": 2740, | |
| "valid_targets_mean": 2196.1, | |
| "valid_targets_min": 1227 | |
| }, | |
| { | |
| "epoch": 4.37799043062201, | |
| "grad_norm": 0.6029868584965588, | |
| "learning_rate": 1.4810816246465374e-05, | |
| "loss": 0.1416, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15540479123592377, | |
| "step": 2745, | |
| "valid_targets_mean": 2765.9, | |
| "valid_targets_min": 1239 | |
| }, | |
| { | |
| "epoch": 4.385964912280702, | |
| "grad_norm": 0.5597027272525674, | |
| "learning_rate": 1.4734047204107358e-05, | |
| "loss": 0.1504, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13575266301631927, | |
| "step": 2750, | |
| "valid_targets_mean": 2654.2, | |
| "valid_targets_min": 1265 | |
| }, | |
| { | |
| "epoch": 4.393939393939394, | |
| "grad_norm": 0.5407528721459675, | |
| "learning_rate": 1.4657361438151192e-05, | |
| "loss": 0.1715, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16623978316783905, | |
| "step": 2755, | |
| "valid_targets_mean": 3222.1, | |
| "valid_targets_min": 1279 | |
| }, | |
| { | |
| "epoch": 4.401913875598086, | |
| "grad_norm": 0.6874305661080543, | |
| "learning_rate": 1.458076016131468e-05, | |
| "loss": 0.1718, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1757175326347351, | |
| "step": 2760, | |
| "valid_targets_mean": 2338.8, | |
| "valid_targets_min": 680 | |
| }, | |
| { | |
| "epoch": 4.409888357256778, | |
| "grad_norm": 0.5591760680100331, | |
| "learning_rate": 1.4504244584979494e-05, | |
| "loss": 0.1568, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17732250690460205, | |
| "step": 2765, | |
| "valid_targets_mean": 3493.1, | |
| "valid_targets_min": 1162 | |
| }, | |
| { | |
| "epoch": 4.417862838915471, | |
| "grad_norm": 0.6006214490581323, | |
| "learning_rate": 1.442781591917203e-05, | |
| "loss": 0.1556, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10994596034288406, | |
| "step": 2770, | |
| "valid_targets_mean": 1990.3, | |
| "valid_targets_min": 1076 | |
| }, | |
| { | |
| "epoch": 4.425837320574162, | |
| "grad_norm": 0.6136597184600974, | |
| "learning_rate": 1.4351475372544262e-05, | |
| "loss": 0.1656, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1602148413658142, | |
| "step": 2775, | |
| "valid_targets_mean": 2516.6, | |
| "valid_targets_min": 1300 | |
| }, | |
| { | |
| "epoch": 4.433811802232855, | |
| "grad_norm": 0.6063956457279249, | |
| "learning_rate": 1.4275224152354658e-05, | |
| "loss": 0.1623, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20220395922660828, | |
| "step": 2780, | |
| "valid_targets_mean": 3064.8, | |
| "valid_targets_min": 1065 | |
| }, | |
| { | |
| "epoch": 4.441786283891547, | |
| "grad_norm": 0.6960689470943858, | |
| "learning_rate": 1.4199063464449047e-05, | |
| "loss": 0.1725, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.23058322072029114, | |
| "step": 2785, | |
| "valid_targets_mean": 2944.1, | |
| "valid_targets_min": 1199 | |
| }, | |
| { | |
| "epoch": 4.44976076555024, | |
| "grad_norm": 0.581843784735055, | |
| "learning_rate": 1.4122994513241587e-05, | |
| "loss": 0.1682, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16655464470386505, | |
| "step": 2790, | |
| "valid_targets_mean": 2789.2, | |
| "valid_targets_min": 1179 | |
| }, | |
| { | |
| "epoch": 4.457735247208931, | |
| "grad_norm": 0.5850624942481338, | |
| "learning_rate": 1.4047018501695677e-05, | |
| "loss": 0.1647, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18775057792663574, | |
| "step": 2795, | |
| "valid_targets_mean": 3112.1, | |
| "valid_targets_min": 978 | |
| }, | |
| { | |
| "epoch": 4.4657097288676235, | |
| "grad_norm": 0.7725419945300261, | |
| "learning_rate": 1.3971136631304978e-05, | |
| "loss": 0.1638, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13948987424373627, | |
| "step": 2800, | |
| "valid_targets_mean": 3131.8, | |
| "valid_targets_min": 1342 | |
| }, | |
| { | |
| "epoch": 4.473684210526316, | |
| "grad_norm": 0.6442207373438252, | |
| "learning_rate": 1.3895350102074386e-05, | |
| "loss": 0.1533, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13303601741790771, | |
| "step": 2805, | |
| "valid_targets_mean": 2095.6, | |
| "valid_targets_min": 876 | |
| }, | |
| { | |
| "epoch": 4.481658692185008, | |
| "grad_norm": 0.6268169440228296, | |
| "learning_rate": 1.3819660112501054e-05, | |
| "loss": 0.1599, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19091369211673737, | |
| "step": 2810, | |
| "valid_targets_mean": 2760.9, | |
| "valid_targets_min": 1127 | |
| }, | |
| { | |
| "epoch": 4.4896331738437, | |
| "grad_norm": 0.7797827360900279, | |
| "learning_rate": 1.3744067859555461e-05, | |
| "loss": 0.1596, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14364150166511536, | |
| "step": 2815, | |
| "valid_targets_mean": 2647.5, | |
| "valid_targets_min": 987 | |
| }, | |
| { | |
| "epoch": 4.497607655502392, | |
| "grad_norm": 0.5958486269846882, | |
| "learning_rate": 1.3668574538662451e-05, | |
| "loss": 0.1908, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3153199553489685, | |
| "step": 2820, | |
| "valid_targets_mean": 4524.3, | |
| "valid_targets_min": 1101 | |
| }, | |
| { | |
| "epoch": 4.505582137161085, | |
| "grad_norm": 0.6464530618817084, | |
| "learning_rate": 1.3593181343682353e-05, | |
| "loss": 0.1509, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12288454174995422, | |
| "step": 2825, | |
| "valid_targets_mean": 1907.5, | |
| "valid_targets_min": 862 | |
| }, | |
| { | |
| "epoch": 4.513556618819777, | |
| "grad_norm": 0.5766089488256633, | |
| "learning_rate": 1.3517889466892085e-05, | |
| "loss": 0.1511, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18093323707580566, | |
| "step": 2830, | |
| "valid_targets_mean": 3210.1, | |
| "valid_targets_min": 1257 | |
| }, | |
| { | |
| "epoch": 4.521531100478469, | |
| "grad_norm": 0.6277488852884846, | |
| "learning_rate": 1.3442700098966326e-05, | |
| "loss": 0.1548, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14466340839862823, | |
| "step": 2835, | |
| "valid_targets_mean": 2295.5, | |
| "valid_targets_min": 1121 | |
| }, | |
| { | |
| "epoch": 4.529505582137161, | |
| "grad_norm": 0.5780933276759791, | |
| "learning_rate": 1.3367614428958646e-05, | |
| "loss": 0.1721, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13822069764137268, | |
| "step": 2840, | |
| "valid_targets_mean": 2436.6, | |
| "valid_targets_min": 1185 | |
| }, | |
| { | |
| "epoch": 4.5374800637958534, | |
| "grad_norm": 0.5182007147664368, | |
| "learning_rate": 1.3292633644282735e-05, | |
| "loss": 0.185, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1259920299053192, | |
| "step": 2845, | |
| "valid_targets_mean": 3187.9, | |
| "valid_targets_min": 1241 | |
| }, | |
| { | |
| "epoch": 4.545454545454545, | |
| "grad_norm": 0.5371658448891603, | |
| "learning_rate": 1.3217758930693608e-05, | |
| "loss": 0.1566, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12785111367702484, | |
| "step": 2850, | |
| "valid_targets_mean": 2966.3, | |
| "valid_targets_min": 1333 | |
| }, | |
| { | |
| "epoch": 4.553429027113237, | |
| "grad_norm": 0.6492823713058845, | |
| "learning_rate": 1.314299147226887e-05, | |
| "loss": 0.1734, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17848286032676697, | |
| "step": 2855, | |
| "valid_targets_mean": 3112.8, | |
| "valid_targets_min": 991 | |
| }, | |
| { | |
| "epoch": 4.56140350877193, | |
| "grad_norm": 0.5833526819664339, | |
| "learning_rate": 1.3068332451389969e-05, | |
| "loss": 0.1619, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14901649951934814, | |
| "step": 2860, | |
| "valid_targets_mean": 2962.1, | |
| "valid_targets_min": 921 | |
| }, | |
| { | |
| "epoch": 4.569377990430622, | |
| "grad_norm": 0.6044763831376962, | |
| "learning_rate": 1.2993783048723515e-05, | |
| "loss": 0.173, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18289071321487427, | |
| "step": 2865, | |
| "valid_targets_mean": 2793.9, | |
| "valid_targets_min": 1306 | |
| }, | |
| { | |
| "epoch": 4.577352472089315, | |
| "grad_norm": 0.5907034874662462, | |
| "learning_rate": 1.2919344443202602e-05, | |
| "loss": 0.1718, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1798447072505951, | |
| "step": 2870, | |
| "valid_targets_mean": 3851.8, | |
| "valid_targets_min": 1072 | |
| }, | |
| { | |
| "epoch": 4.585326953748006, | |
| "grad_norm": 0.5614296037672958, | |
| "learning_rate": 1.2845017812008158e-05, | |
| "loss": 0.1686, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1698157787322998, | |
| "step": 2875, | |
| "valid_targets_mean": 3441.5, | |
| "valid_targets_min": 1223 | |
| }, | |
| { | |
| "epoch": 4.5933014354066986, | |
| "grad_norm": 0.5713076100304711, | |
| "learning_rate": 1.277080433055034e-05, | |
| "loss": 0.1648, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14279823005199432, | |
| "step": 2880, | |
| "valid_targets_mean": 2656.8, | |
| "valid_targets_min": 1218 | |
| }, | |
| { | |
| "epoch": 4.601275917065391, | |
| "grad_norm": 0.6010777119114208, | |
| "learning_rate": 1.2696705172449944e-05, | |
| "loss": 0.1716, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16096925735473633, | |
| "step": 2885, | |
| "valid_targets_mean": 2882.9, | |
| "valid_targets_min": 1236 | |
| }, | |
| { | |
| "epoch": 4.6092503987240825, | |
| "grad_norm": 0.6510728768592932, | |
| "learning_rate": 1.2622721509519846e-05, | |
| "loss": 0.2013, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18414732813835144, | |
| "step": 2890, | |
| "valid_targets_mean": 2611.4, | |
| "valid_targets_min": 1102 | |
| }, | |
| { | |
| "epoch": 4.617224880382775, | |
| "grad_norm": 0.5891857819758494, | |
| "learning_rate": 1.2548854511746461e-05, | |
| "loss": 0.2005, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1432403326034546, | |
| "step": 2895, | |
| "valid_targets_mean": 2906.4, | |
| "valid_targets_min": 1053 | |
| }, | |
| { | |
| "epoch": 4.625199362041467, | |
| "grad_norm": 0.5997581297495093, | |
| "learning_rate": 1.247510534727125e-05, | |
| "loss": 0.1722, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17660380899906158, | |
| "step": 2900, | |
| "valid_targets_mean": 4153.6, | |
| "valid_targets_min": 1566 | |
| }, | |
| { | |
| "epoch": 4.63317384370016, | |
| "grad_norm": 0.5704940600430232, | |
| "learning_rate": 1.240147518237224e-05, | |
| "loss": 0.1698, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17439067363739014, | |
| "step": 2905, | |
| "valid_targets_mean": 2826.9, | |
| "valid_targets_min": 1557 | |
| }, | |
| { | |
| "epoch": 4.641148325358852, | |
| "grad_norm": 0.6041235814136545, | |
| "learning_rate": 1.2327965181445593e-05, | |
| "loss": 0.1612, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1622115969657898, | |
| "step": 2910, | |
| "valid_targets_mean": 2654.2, | |
| "valid_targets_min": 1260 | |
| }, | |
| { | |
| "epoch": 4.649122807017544, | |
| "grad_norm": 0.5904064735068515, | |
| "learning_rate": 1.2254576506987182e-05, | |
| "loss": 0.1571, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14826351404190063, | |
| "step": 2915, | |
| "valid_targets_mean": 2521.2, | |
| "valid_targets_min": 1203 | |
| }, | |
| { | |
| "epoch": 4.657097288676236, | |
| "grad_norm": 0.6708883356957397, | |
| "learning_rate": 1.21813103195742e-05, | |
| "loss": 0.1874, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15236374735832214, | |
| "step": 2920, | |
| "valid_targets_mean": 2082.2, | |
| "valid_targets_min": 1310 | |
| }, | |
| { | |
| "epoch": 4.6650717703349285, | |
| "grad_norm": 0.5410250981706312, | |
| "learning_rate": 1.2108167777846815e-05, | |
| "loss": 0.1647, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1691250205039978, | |
| "step": 2925, | |
| "valid_targets_mean": 3658.4, | |
| "valid_targets_min": 1119 | |
| }, | |
| { | |
| "epoch": 4.67304625199362, | |
| "grad_norm": 0.5600363452922104, | |
| "learning_rate": 1.203515003848987e-05, | |
| "loss": 0.1532, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17247727513313293, | |
| "step": 2930, | |
| "valid_targets_mean": 3237.1, | |
| "valid_targets_min": 1034 | |
| }, | |
| { | |
| "epoch": 4.681020733652312, | |
| "grad_norm": 0.6269773523410778, | |
| "learning_rate": 1.1962258256214545e-05, | |
| "loss": 0.1553, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16801588237285614, | |
| "step": 2935, | |
| "valid_targets_mean": 2773.1, | |
| "valid_targets_min": 1145 | |
| }, | |
| { | |
| "epoch": 4.688995215311005, | |
| "grad_norm": 0.6416942531641958, | |
| "learning_rate": 1.1889493583740124e-05, | |
| "loss": 0.1488, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15715159475803375, | |
| "step": 2940, | |
| "valid_targets_mean": 2487.7, | |
| "valid_targets_min": 925 | |
| }, | |
| { | |
| "epoch": 4.696969696969697, | |
| "grad_norm": 0.6032413565847697, | |
| "learning_rate": 1.1816857171775767e-05, | |
| "loss": 0.1401, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15060892701148987, | |
| "step": 2945, | |
| "valid_targets_mean": 2569.9, | |
| "valid_targets_min": 1571 | |
| }, | |
| { | |
| "epoch": 4.70494417862839, | |
| "grad_norm": 0.6229235355749451, | |
| "learning_rate": 1.1744350169002308e-05, | |
| "loss": 0.1727, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16957922279834747, | |
| "step": 2950, | |
| "valid_targets_mean": 2637.2, | |
| "valid_targets_min": 1252 | |
| }, | |
| { | |
| "epoch": 4.712918660287081, | |
| "grad_norm": 0.5305233008927874, | |
| "learning_rate": 1.1671973722054077e-05, | |
| "loss": 0.1707, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20151448249816895, | |
| "step": 2955, | |
| "valid_targets_mean": 3813.4, | |
| "valid_targets_min": 952 | |
| }, | |
| { | |
| "epoch": 4.720893141945774, | |
| "grad_norm": 0.5928610616782457, | |
| "learning_rate": 1.159972897550079e-05, | |
| "loss": 0.1765, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17702104151248932, | |
| "step": 2960, | |
| "valid_targets_mean": 3135.0, | |
| "valid_targets_min": 1101 | |
| }, | |
| { | |
| "epoch": 4.728867623604466, | |
| "grad_norm": 0.6549965052438604, | |
| "learning_rate": 1.1527617071829447e-05, | |
| "loss": 0.2088, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21487164497375488, | |
| "step": 2965, | |
| "valid_targets_mean": 2599.4, | |
| "valid_targets_min": 1431 | |
| }, | |
| { | |
| "epoch": 4.7368421052631575, | |
| "grad_norm": 0.5878382112347069, | |
| "learning_rate": 1.145563915142622e-05, | |
| "loss": 0.1647, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16524244844913483, | |
| "step": 2970, | |
| "valid_targets_mean": 3205.4, | |
| "valid_targets_min": 753 | |
| }, | |
| { | |
| "epoch": 4.74481658692185, | |
| "grad_norm": 0.5829785508403137, | |
| "learning_rate": 1.1383796352558489e-05, | |
| "loss": 0.16, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13480517268180847, | |
| "step": 2975, | |
| "valid_targets_mean": 2421.6, | |
| "valid_targets_min": 1389 | |
| }, | |
| { | |
| "epoch": 4.752791068580542, | |
| "grad_norm": 0.9528420577486767, | |
| "learning_rate": 1.1312089811356803e-05, | |
| "loss": 0.1511, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.142283096909523, | |
| "step": 2980, | |
| "valid_targets_mean": 2265.3, | |
| "valid_targets_min": 1130 | |
| }, | |
| { | |
| "epoch": 4.760765550239235, | |
| "grad_norm": 0.6379187971048914, | |
| "learning_rate": 1.1240520661796882e-05, | |
| "loss": 0.1434, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12618593871593475, | |
| "step": 2985, | |
| "valid_targets_mean": 2114.1, | |
| "valid_targets_min": 971 | |
| }, | |
| { | |
| "epoch": 4.768740031897926, | |
| "grad_norm": 0.581096748192637, | |
| "learning_rate": 1.1169090035681772e-05, | |
| "loss": 0.1541, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17308899760246277, | |
| "step": 2990, | |
| "valid_targets_mean": 3163.0, | |
| "valid_targets_min": 1331 | |
| }, | |
| { | |
| "epoch": 4.776714513556619, | |
| "grad_norm": 0.6490837955490153, | |
| "learning_rate": 1.109779906262383e-05, | |
| "loss": 0.1918, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2551341652870178, | |
| "step": 2995, | |
| "valid_targets_mean": 3147.7, | |
| "valid_targets_min": 1303 | |
| }, | |
| { | |
| "epoch": 4.784688995215311, | |
| "grad_norm": 0.5933551257310504, | |
| "learning_rate": 1.102664887002698e-05, | |
| "loss": 0.1725, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15214470028877258, | |
| "step": 3000, | |
| "valid_targets_mean": 3350.6, | |
| "valid_targets_min": 1496 | |
| }, | |
| { | |
| "epoch": 4.7926634768740035, | |
| "grad_norm": 0.5823345389450942, | |
| "learning_rate": 1.0955640583068802e-05, | |
| "loss": 0.1512, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16861368715763092, | |
| "step": 3005, | |
| "valid_targets_mean": 2921.8, | |
| "valid_targets_min": 1039 | |
| }, | |
| { | |
| "epoch": 4.800637958532695, | |
| "grad_norm": 0.5903270104669966, | |
| "learning_rate": 1.0884775324682755e-05, | |
| "loss": 0.1625, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14034119248390198, | |
| "step": 3010, | |
| "valid_targets_mean": 2480.7, | |
| "valid_targets_min": 998 | |
| }, | |
| { | |
| "epoch": 4.8086124401913874, | |
| "grad_norm": 0.5924958725908289, | |
| "learning_rate": 1.081405421554044e-05, | |
| "loss": 0.1676, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13198724389076233, | |
| "step": 3015, | |
| "valid_targets_mean": 2427.5, | |
| "valid_targets_min": 1250 | |
| }, | |
| { | |
| "epoch": 4.81658692185008, | |
| "grad_norm": 0.5951560445307537, | |
| "learning_rate": 1.074347837403387e-05, | |
| "loss": 0.1797, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1821703463792801, | |
| "step": 3020, | |
| "valid_targets_mean": 3043.9, | |
| "valid_targets_min": 1239 | |
| }, | |
| { | |
| "epoch": 4.824561403508772, | |
| "grad_norm": 0.5894961447208159, | |
| "learning_rate": 1.067304891625776e-05, | |
| "loss": 0.1517, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1578672230243683, | |
| "step": 3025, | |
| "valid_targets_mean": 3003.2, | |
| "valid_targets_min": 1063 | |
| }, | |
| { | |
| "epoch": 4.832535885167464, | |
| "grad_norm": 0.6377603875579224, | |
| "learning_rate": 1.0602766955991913e-05, | |
| "loss": 0.1634, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16185033321380615, | |
| "step": 3030, | |
| "valid_targets_mean": 2483.1, | |
| "valid_targets_min": 1017 | |
| }, | |
| { | |
| "epoch": 4.840510366826156, | |
| "grad_norm": 0.48915397672553124, | |
| "learning_rate": 1.053263360468358e-05, | |
| "loss": 0.1749, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1294403374195099, | |
| "step": 3035, | |
| "valid_targets_mean": 3507.8, | |
| "valid_targets_min": 1335 | |
| }, | |
| { | |
| "epoch": 4.848484848484849, | |
| "grad_norm": 0.6547180933536588, | |
| "learning_rate": 1.0462649971429884e-05, | |
| "loss": 0.1794, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1606912463903427, | |
| "step": 3040, | |
| "valid_targets_mean": 2161.8, | |
| "valid_targets_min": 670 | |
| }, | |
| { | |
| "epoch": 4.856459330143541, | |
| "grad_norm": 0.5527293462604799, | |
| "learning_rate": 1.0392817162960304e-05, | |
| "loss": 0.1585, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17694956064224243, | |
| "step": 3045, | |
| "valid_targets_mean": 3561.0, | |
| "valid_targets_min": 1423 | |
| }, | |
| { | |
| "epoch": 4.8644338118022326, | |
| "grad_norm": 0.7548532786478716, | |
| "learning_rate": 1.0323136283619167e-05, | |
| "loss": 0.1678, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15897555649280548, | |
| "step": 3050, | |
| "valid_targets_mean": 2484.4, | |
| "valid_targets_min": 1075 | |
| }, | |
| { | |
| "epoch": 4.872408293460925, | |
| "grad_norm": 0.5513293334710877, | |
| "learning_rate": 1.0253608435348136e-05, | |
| "loss": 0.1612, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18371330201625824, | |
| "step": 3055, | |
| "valid_targets_mean": 3966.9, | |
| "valid_targets_min": 1121 | |
| }, | |
| { | |
| "epoch": 4.880382775119617, | |
| "grad_norm": 0.6031939961843086, | |
| "learning_rate": 1.0184234717668867e-05, | |
| "loss": 0.1702, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18013069033622742, | |
| "step": 3060, | |
| "valid_targets_mean": 2705.5, | |
| "valid_targets_min": 1237 | |
| }, | |
| { | |
| "epoch": 4.88835725677831, | |
| "grad_norm": 0.6892842856879766, | |
| "learning_rate": 1.0115016227665544e-05, | |
| "loss": 0.1366, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13218329846858978, | |
| "step": 3065, | |
| "valid_targets_mean": 1881.2, | |
| "valid_targets_min": 1069 | |
| }, | |
| { | |
| "epoch": 4.896331738437001, | |
| "grad_norm": 0.543825197475642, | |
| "learning_rate": 1.0045954059967577e-05, | |
| "loss": 0.1766, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13563859462738037, | |
| "step": 3070, | |
| "valid_targets_mean": 3105.5, | |
| "valid_targets_min": 1430 | |
| }, | |
| { | |
| "epoch": 4.904306220095694, | |
| "grad_norm": 0.6591944511943134, | |
| "learning_rate": 9.977049306732287e-06, | |
| "loss": 0.1695, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15450099110603333, | |
| "step": 3075, | |
| "valid_targets_mean": 2669.7, | |
| "valid_targets_min": 973 | |
| }, | |
| { | |
| "epoch": 4.912280701754386, | |
| "grad_norm": 0.5795314772456565, | |
| "learning_rate": 9.908303057627591e-06, | |
| "loss": 0.1787, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16409224271774292, | |
| "step": 3080, | |
| "valid_targets_mean": 3158.5, | |
| "valid_targets_min": 1344 | |
| }, | |
| { | |
| "epoch": 4.920255183413078, | |
| "grad_norm": 0.6592054493281521, | |
| "learning_rate": 9.83971639981484e-06, | |
| "loss": 0.1901, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.22291454672813416, | |
| "step": 3085, | |
| "valid_targets_mean": 3663.3, | |
| "valid_targets_min": 1271 | |
| }, | |
| { | |
| "epoch": 4.92822966507177, | |
| "grad_norm": 0.5726489277415606, | |
| "learning_rate": 9.771290417931559e-06, | |
| "loss": 0.1557, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16419470310211182, | |
| "step": 3090, | |
| "valid_targets_mean": 2881.9, | |
| "valid_targets_min": 1361 | |
| }, | |
| { | |
| "epoch": 4.9362041467304625, | |
| "grad_norm": 0.6137702893297173, | |
| "learning_rate": 9.703026194074342e-06, | |
| "loss": 0.1597, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16602787375450134, | |
| "step": 3095, | |
| "valid_targets_mean": 2241.7, | |
| "valid_targets_min": 1492 | |
| }, | |
| { | |
| "epoch": 4.944178628389155, | |
| "grad_norm": 0.625860695726005, | |
| "learning_rate": 9.634924807781729e-06, | |
| "loss": 0.1575, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13178279995918274, | |
| "step": 3100, | |
| "valid_targets_mean": 2148.7, | |
| "valid_targets_min": 1197 | |
| }, | |
| { | |
| "epoch": 4.952153110047847, | |
| "grad_norm": 0.6614163326110132, | |
| "learning_rate": 9.566987336017102e-06, | |
| "loss": 0.1518, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15211498737335205, | |
| "step": 3105, | |
| "valid_targets_mean": 2294.9, | |
| "valid_targets_min": 1449 | |
| }, | |
| { | |
| "epoch": 4.960127591706539, | |
| "grad_norm": 0.6339096958861727, | |
| "learning_rate": 9.499214853151699e-06, | |
| "loss": 0.1642, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13669738173484802, | |
| "step": 3110, | |
| "valid_targets_mean": 2579.0, | |
| "valid_targets_min": 1200 | |
| }, | |
| { | |
| "epoch": 4.968102073365231, | |
| "grad_norm": 0.6201778623086089, | |
| "learning_rate": 9.431608430947619e-06, | |
| "loss": 0.1707, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13577091693878174, | |
| "step": 3115, | |
| "valid_targets_mean": 2413.5, | |
| "valid_targets_min": 1018 | |
| }, | |
| { | |
| "epoch": 4.976076555023924, | |
| "grad_norm": 0.5945194233107883, | |
| "learning_rate": 9.364169138540805e-06, | |
| "loss": 0.1741, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16690418124198914, | |
| "step": 3120, | |
| "valid_targets_mean": 2853.1, | |
| "valid_targets_min": 1055 | |
| }, | |
| { | |
| "epoch": 4.984051036682615, | |
| "grad_norm": 0.583377612270226, | |
| "learning_rate": 9.296898042424237e-06, | |
| "loss": 0.1659, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13808023929595947, | |
| "step": 3125, | |
| "valid_targets_mean": 2496.2, | |
| "valid_targets_min": 986 | |
| }, | |
| { | |
| "epoch": 4.992025518341308, | |
| "grad_norm": 0.5532057544654067, | |
| "learning_rate": 9.229796206431015e-06, | |
| "loss": 0.1628, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2222648710012436, | |
| "step": 3130, | |
| "valid_targets_mean": 4848.7, | |
| "valid_targets_min": 1038 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 0.531789644788238, | |
| "learning_rate": 9.162864691717513e-06, | |
| "loss": 0.171, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18254606425762177, | |
| "step": 3135, | |
| "valid_targets_mean": 3269.1, | |
| "valid_targets_min": 1077 | |
| }, | |
| { | |
| "epoch": 5.007974481658692, | |
| "grad_norm": 0.5616721527680314, | |
| "learning_rate": 9.096104556746654e-06, | |
| "loss": 0.1597, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1832147091627121, | |
| "step": 3140, | |
| "valid_targets_mean": 3542.6, | |
| "valid_targets_min": 1594 | |
| }, | |
| { | |
| "epoch": 5.015948963317384, | |
| "grad_norm": 0.6088692960009601, | |
| "learning_rate": 9.029516857271115e-06, | |
| "loss": 0.146, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1366785764694214, | |
| "step": 3145, | |
| "valid_targets_mean": 2708.7, | |
| "valid_targets_min": 1023 | |
| }, | |
| { | |
| "epoch": 5.023923444976076, | |
| "grad_norm": 0.5285306796042663, | |
| "learning_rate": 8.963102646316677e-06, | |
| "loss": 0.1594, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16185930371284485, | |
| "step": 3150, | |
| "valid_targets_mean": 3778.3, | |
| "valid_targets_min": 1154 | |
| }, | |
| { | |
| "epoch": 5.031897926634769, | |
| "grad_norm": 0.6115365337267124, | |
| "learning_rate": 8.896862974165553e-06, | |
| "loss": 0.1478, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14206919074058533, | |
| "step": 3155, | |
| "valid_targets_mean": 3075.5, | |
| "valid_targets_min": 1096 | |
| }, | |
| { | |
| "epoch": 5.039872408293461, | |
| "grad_norm": 0.6388639498863427, | |
| "learning_rate": 8.830798888339756e-06, | |
| "loss": 0.1512, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13919368386268616, | |
| "step": 3160, | |
| "valid_targets_mean": 2328.9, | |
| "valid_targets_min": 1037 | |
| }, | |
| { | |
| "epoch": 5.047846889952153, | |
| "grad_norm": 0.5869360038685693, | |
| "learning_rate": 8.764911433584581e-06, | |
| "loss": 0.1635, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17949020862579346, | |
| "step": 3165, | |
| "valid_targets_mean": 3294.6, | |
| "valid_targets_min": 907 | |
| }, | |
| { | |
| "epoch": 5.055821371610845, | |
| "grad_norm": 0.5747885813337419, | |
| "learning_rate": 8.699201651852056e-06, | |
| "loss": 0.1382, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15749900043010712, | |
| "step": 3170, | |
| "valid_targets_mean": 3640.9, | |
| "valid_targets_min": 1226 | |
| }, | |
| { | |
| "epoch": 5.0637958532695375, | |
| "grad_norm": 0.5817079808666326, | |
| "learning_rate": 8.633670582284446e-06, | |
| "loss": 0.1524, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11290596425533295, | |
| "step": 3175, | |
| "valid_targets_mean": 2822.5, | |
| "valid_targets_min": 804 | |
| }, | |
| { | |
| "epoch": 5.07177033492823, | |
| "grad_norm": 0.6680362984019289, | |
| "learning_rate": 8.56831926119787e-06, | |
| "loss": 0.1423, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11178924143314362, | |
| "step": 3180, | |
| "valid_targets_mean": 1960.3, | |
| "valid_targets_min": 904 | |
| }, | |
| { | |
| "epoch": 5.0797448165869215, | |
| "grad_norm": 0.584769533808893, | |
| "learning_rate": 8.503148722065851e-06, | |
| "loss": 0.1894, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2956717610359192, | |
| "step": 3185, | |
| "valid_targets_mean": 4416.9, | |
| "valid_targets_min": 1142 | |
| }, | |
| { | |
| "epoch": 5.087719298245614, | |
| "grad_norm": 0.572702746875333, | |
| "learning_rate": 8.43815999550303e-06, | |
| "loss": 0.1586, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1503574550151825, | |
| "step": 3190, | |
| "valid_targets_mean": 3477.9, | |
| "valid_targets_min": 1398 | |
| }, | |
| { | |
| "epoch": 5.095693779904306, | |
| "grad_norm": 0.6379898231385502, | |
| "learning_rate": 8.373354109248842e-06, | |
| "loss": 0.1534, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15601392090320587, | |
| "step": 3195, | |
| "valid_targets_mean": 2786.6, | |
| "valid_targets_min": 1274 | |
| }, | |
| { | |
| "epoch": 5.103668261562999, | |
| "grad_norm": 0.6570400100898655, | |
| "learning_rate": 8.308732088151245e-06, | |
| "loss": 0.1837, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13875964283943176, | |
| "step": 3200, | |
| "valid_targets_mean": 2495.5, | |
| "valid_targets_min": 1253 | |
| }, | |
| { | |
| "epoch": 5.11164274322169, | |
| "grad_norm": 0.598589916212468, | |
| "learning_rate": 8.24429495415054e-06, | |
| "loss": 0.1542, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17209793627262115, | |
| "step": 3205, | |
| "valid_targets_mean": 3544.8, | |
| "valid_targets_min": 1088 | |
| }, | |
| { | |
| "epoch": 5.119617224880383, | |
| "grad_norm": 0.5905187204371102, | |
| "learning_rate": 8.180043726263216e-06, | |
| "loss": 0.1563, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15271854400634766, | |
| "step": 3210, | |
| "valid_targets_mean": 3155.5, | |
| "valid_targets_min": 1290 | |
| }, | |
| { | |
| "epoch": 5.127591706539075, | |
| "grad_norm": 0.638961162391656, | |
| "learning_rate": 8.115979420565794e-06, | |
| "loss": 0.1467, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15162287652492523, | |
| "step": 3215, | |
| "valid_targets_mean": 3120.7, | |
| "valid_targets_min": 1046 | |
| }, | |
| { | |
| "epoch": 5.1355661881977674, | |
| "grad_norm": 0.6929090202401988, | |
| "learning_rate": 8.052103050178806e-06, | |
| "loss": 0.1451, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15893588960170746, | |
| "step": 3220, | |
| "valid_targets_mean": 2715.9, | |
| "valid_targets_min": 1185 | |
| }, | |
| { | |
| "epoch": 5.143540669856459, | |
| "grad_norm": 0.6531239078174592, | |
| "learning_rate": 7.988415625250755e-06, | |
| "loss": 0.1755, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17716528475284576, | |
| "step": 3225, | |
| "valid_targets_mean": 3181.6, | |
| "valid_targets_min": 1121 | |
| }, | |
| { | |
| "epoch": 5.151515151515151, | |
| "grad_norm": 0.6723920338293383, | |
| "learning_rate": 7.924918152942117e-06, | |
| "loss": 0.146, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16147425770759583, | |
| "step": 3230, | |
| "valid_targets_mean": 2302.5, | |
| "valid_targets_min": 1238 | |
| }, | |
| { | |
| "epoch": 5.159489633173844, | |
| "grad_norm": 0.6602533766749181, | |
| "learning_rate": 7.861611637409462e-06, | |
| "loss": 0.1376, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1290948987007141, | |
| "step": 3235, | |
| "valid_targets_mean": 1941.5, | |
| "valid_targets_min": 947 | |
| }, | |
| { | |
| "epoch": 5.167464114832536, | |
| "grad_norm": 0.6264465102021803, | |
| "learning_rate": 7.798497079789513e-06, | |
| "loss": 0.1616, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16076138615608215, | |
| "step": 3240, | |
| "valid_targets_mean": 3415.3, | |
| "valid_targets_min": 1347 | |
| }, | |
| { | |
| "epoch": 5.175438596491228, | |
| "grad_norm": 0.6240803701984488, | |
| "learning_rate": 7.735575478183381e-06, | |
| "loss": 0.1501, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11426462233066559, | |
| "step": 3245, | |
| "valid_targets_mean": 2283.6, | |
| "valid_targets_min": 1214 | |
| }, | |
| { | |
| "epoch": 5.18341307814992, | |
| "grad_norm": 0.5078898685113183, | |
| "learning_rate": 7.672847827640735e-06, | |
| "loss": 0.1382, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10710498690605164, | |
| "step": 3250, | |
| "valid_targets_mean": 3287.6, | |
| "valid_targets_min": 1326 | |
| }, | |
| { | |
| "epoch": 5.1913875598086126, | |
| "grad_norm": 0.6613642718503813, | |
| "learning_rate": 7.610315120144067e-06, | |
| "loss": 0.1686, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2189890742301941, | |
| "step": 3255, | |
| "valid_targets_mean": 3137.3, | |
| "valid_targets_min": 1337 | |
| }, | |
| { | |
| "epoch": 5.199362041467305, | |
| "grad_norm": 0.6542926532200951, | |
| "learning_rate": 7.5479783445930414e-06, | |
| "loss": 0.1436, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13109362125396729, | |
| "step": 3260, | |
| "valid_targets_mean": 2605.9, | |
| "valid_targets_min": 1169 | |
| }, | |
| { | |
| "epoch": 5.2073365231259965, | |
| "grad_norm": 0.6132656516318115, | |
| "learning_rate": 7.485838486788803e-06, | |
| "loss": 0.1566, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1521100401878357, | |
| "step": 3265, | |
| "valid_targets_mean": 2872.9, | |
| "valid_targets_min": 1194 | |
| }, | |
| { | |
| "epoch": 5.215311004784689, | |
| "grad_norm": 0.5793143995524747, | |
| "learning_rate": 7.4238965294184374e-06, | |
| "loss": 0.1398, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1307733952999115, | |
| "step": 3270, | |
| "valid_targets_mean": 2821.3, | |
| "valid_targets_min": 1036 | |
| }, | |
| { | |
| "epoch": 5.223285486443381, | |
| "grad_norm": 0.616310991142214, | |
| "learning_rate": 7.362153452039409e-06, | |
| "loss": 0.1468, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14505207538604736, | |
| "step": 3275, | |
| "valid_targets_mean": 3195.2, | |
| "valid_targets_min": 906 | |
| }, | |
| { | |
| "epoch": 5.231259968102074, | |
| "grad_norm": 0.6618556740317185, | |
| "learning_rate": 7.300610231064056e-06, | |
| "loss": 0.1609, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1720818281173706, | |
| "step": 3280, | |
| "valid_targets_mean": 2992.6, | |
| "valid_targets_min": 1399 | |
| }, | |
| { | |
| "epoch": 5.239234449760765, | |
| "grad_norm": 0.5790679579947826, | |
| "learning_rate": 7.239267839744166e-06, | |
| "loss": 0.1538, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14833727478981018, | |
| "step": 3285, | |
| "valid_targets_mean": 3526.7, | |
| "valid_targets_min": 1427 | |
| }, | |
| { | |
| "epoch": 5.247208931419458, | |
| "grad_norm": 0.6637878617650812, | |
| "learning_rate": 7.178127248155604e-06, | |
| "loss": 0.1422, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14050063490867615, | |
| "step": 3290, | |
| "valid_targets_mean": 2340.9, | |
| "valid_targets_min": 1305 | |
| }, | |
| { | |
| "epoch": 5.25518341307815, | |
| "grad_norm": 0.660304590549916, | |
| "learning_rate": 7.117189423182917e-06, | |
| "loss": 0.1772, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12989658117294312, | |
| "step": 3295, | |
| "valid_targets_mean": 2483.1, | |
| "valid_targets_min": 1431 | |
| }, | |
| { | |
| "epoch": 5.2631578947368425, | |
| "grad_norm": 0.5943952448901542, | |
| "learning_rate": 7.056455328504104e-06, | |
| "loss": 0.1338, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11984672397375107, | |
| "step": 3300, | |
| "valid_targets_mean": 2756.2, | |
| "valid_targets_min": 1486 | |
| }, | |
| { | |
| "epoch": 5.271132376395534, | |
| "grad_norm": 0.7138160232972223, | |
| "learning_rate": 6.995925924575342e-06, | |
| "loss": 0.1753, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19342780113220215, | |
| "step": 3305, | |
| "valid_targets_mean": 2799.3, | |
| "valid_targets_min": 723 | |
| }, | |
| { | |
| "epoch": 5.279106858054226, | |
| "grad_norm": 0.6073033076078348, | |
| "learning_rate": 6.935602168615792e-06, | |
| "loss": 0.1697, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17981265485286713, | |
| "step": 3310, | |
| "valid_targets_mean": 3221.4, | |
| "valid_targets_min": 1094 | |
| }, | |
| { | |
| "epoch": 5.287081339712919, | |
| "grad_norm": 0.6198408636633029, | |
| "learning_rate": 6.875485014592493e-06, | |
| "loss": 0.1542, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15747490525245667, | |
| "step": 3315, | |
| "valid_targets_mean": 2937.5, | |
| "valid_targets_min": 946 | |
| }, | |
| { | |
| "epoch": 5.295055821371611, | |
| "grad_norm": 0.6767478976464635, | |
| "learning_rate": 6.815575413205235e-06, | |
| "loss": 0.1414, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1285533607006073, | |
| "step": 3320, | |
| "valid_targets_mean": 1891.3, | |
| "valid_targets_min": 1261 | |
| }, | |
| { | |
| "epoch": 5.303030303030303, | |
| "grad_norm": 0.7336904785294537, | |
| "learning_rate": 6.755874311871562e-06, | |
| "loss": 0.1618, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17466244101524353, | |
| "step": 3325, | |
| "valid_targets_mean": 3122.5, | |
| "valid_targets_min": 1476 | |
| }, | |
| { | |
| "epoch": 5.311004784688995, | |
| "grad_norm": 0.6194185101743505, | |
| "learning_rate": 6.696382654711777e-06, | |
| "loss": 0.1411, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17486578226089478, | |
| "step": 3330, | |
| "valid_targets_mean": 3025.5, | |
| "valid_targets_min": 1313 | |
| }, | |
| { | |
| "epoch": 5.318979266347688, | |
| "grad_norm": 0.560663116939223, | |
| "learning_rate": 6.637101382533986e-06, | |
| "loss": 0.1459, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14362575113773346, | |
| "step": 3335, | |
| "valid_targets_mean": 4083.6, | |
| "valid_targets_min": 1130 | |
| }, | |
| { | |
| "epoch": 5.32695374800638, | |
| "grad_norm": 0.558896380159237, | |
| "learning_rate": 6.578031432819263e-06, | |
| "loss": 0.1324, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13198032975196838, | |
| "step": 3340, | |
| "valid_targets_mean": 3352.3, | |
| "valid_targets_min": 1006 | |
| }, | |
| { | |
| "epoch": 5.3349282296650715, | |
| "grad_norm": 0.5773043179265, | |
| "learning_rate": 6.5191737397068015e-06, | |
| "loss": 0.145, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14222127199172974, | |
| "step": 3345, | |
| "valid_targets_mean": 3164.2, | |
| "valid_targets_min": 1122 | |
| }, | |
| { | |
| "epoch": 5.342902711323764, | |
| "grad_norm": 0.6580880391527406, | |
| "learning_rate": 6.460529233979127e-06, | |
| "loss": 0.1575, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14480651915073395, | |
| "step": 3350, | |
| "valid_targets_mean": 2353.7, | |
| "valid_targets_min": 1181 | |
| }, | |
| { | |
| "epoch": 5.350877192982456, | |
| "grad_norm": 0.6843060196894174, | |
| "learning_rate": 6.402098843047417e-06, | |
| "loss": 0.1673, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14741107821464539, | |
| "step": 3355, | |
| "valid_targets_mean": 2423.2, | |
| "valid_targets_min": 1307 | |
| }, | |
| { | |
| "epoch": 5.358851674641148, | |
| "grad_norm": 0.6168049212851636, | |
| "learning_rate": 6.343883490936791e-06, | |
| "loss": 0.152, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19296379387378693, | |
| "step": 3360, | |
| "valid_targets_mean": 2869.9, | |
| "valid_targets_min": 1109 | |
| }, | |
| { | |
| "epoch": 5.36682615629984, | |
| "grad_norm": 0.6000631965540703, | |
| "learning_rate": 6.285884098271739e-06, | |
| "loss": 0.1371, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13257285952568054, | |
| "step": 3365, | |
| "valid_targets_mean": 2764.1, | |
| "valid_targets_min": 1216 | |
| }, | |
| { | |
| "epoch": 5.374800637958533, | |
| "grad_norm": 0.6563286214420883, | |
| "learning_rate": 6.228101582261532e-06, | |
| "loss": 0.156, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17767846584320068, | |
| "step": 3370, | |
| "valid_targets_mean": 2802.8, | |
| "valid_targets_min": 925 | |
| }, | |
| { | |
| "epoch": 5.382775119617225, | |
| "grad_norm": 0.6960377409393284, | |
| "learning_rate": 6.170536856685716e-06, | |
| "loss": 0.1399, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14913170039653778, | |
| "step": 3375, | |
| "valid_targets_mean": 2647.2, | |
| "valid_targets_min": 983 | |
| }, | |
| { | |
| "epoch": 5.3907496012759175, | |
| "grad_norm": 0.6035783453810492, | |
| "learning_rate": 6.113190831879698e-06, | |
| "loss": 0.1538, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1609422266483307, | |
| "step": 3380, | |
| "valid_targets_mean": 2827.8, | |
| "valid_targets_min": 464 | |
| }, | |
| { | |
| "epoch": 5.398724082934609, | |
| "grad_norm": 0.6576832684505226, | |
| "learning_rate": 6.056064414720317e-06, | |
| "loss": 0.1575, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1647460162639618, | |
| "step": 3385, | |
| "valid_targets_mean": 2748.5, | |
| "valid_targets_min": 1202 | |
| }, | |
| { | |
| "epoch": 5.4066985645933014, | |
| "grad_norm": 0.6402347861728089, | |
| "learning_rate": 5.999158508611496e-06, | |
| "loss": 0.1494, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13792529702186584, | |
| "step": 3390, | |
| "valid_targets_mean": 2541.5, | |
| "valid_targets_min": 1040 | |
| }, | |
| { | |
| "epoch": 5.414673046251994, | |
| "grad_norm": 0.6754352948702881, | |
| "learning_rate": 5.942474013469983e-06, | |
| "loss": 0.1415, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12908028066158295, | |
| "step": 3395, | |
| "valid_targets_mean": 2440.5, | |
| "valid_targets_min": 1076 | |
| }, | |
| { | |
| "epoch": 5.422647527910685, | |
| "grad_norm": 0.6188190631444767, | |
| "learning_rate": 5.886011825711117e-06, | |
| "loss": 0.1507, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15597115457057953, | |
| "step": 3400, | |
| "valid_targets_mean": 3018.2, | |
| "valid_targets_min": 1044 | |
| }, | |
| { | |
| "epoch": 5.430622009569378, | |
| "grad_norm": 0.6027543335163618, | |
| "learning_rate": 5.829772838234615e-06, | |
| "loss": 0.1541, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14923419058322906, | |
| "step": 3405, | |
| "valid_targets_mean": 3284.5, | |
| "valid_targets_min": 1227 | |
| }, | |
| { | |
| "epoch": 5.43859649122807, | |
| "grad_norm": 0.6397744296419502, | |
| "learning_rate": 5.773757940410503e-06, | |
| "loss": 0.1446, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1353510320186615, | |
| "step": 3410, | |
| "valid_targets_mean": 2694.5, | |
| "valid_targets_min": 1031 | |
| }, | |
| { | |
| "epoch": 5.446570972886763, | |
| "grad_norm": 0.6602471322717809, | |
| "learning_rate": 5.7179680180650055e-06, | |
| "loss": 0.1592, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12969273328781128, | |
| "step": 3415, | |
| "valid_targets_mean": 2616.6, | |
| "valid_targets_min": 933 | |
| }, | |
| { | |
| "epoch": 5.454545454545454, | |
| "grad_norm": 0.5963165890470659, | |
| "learning_rate": 5.6624039534665775e-06, | |
| "loss": 0.1587, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13266560435295105, | |
| "step": 3420, | |
| "valid_targets_mean": 3448.6, | |
| "valid_targets_min": 1318 | |
| }, | |
| { | |
| "epoch": 5.4625199362041466, | |
| "grad_norm": 0.5800990660141001, | |
| "learning_rate": 5.607066625311925e-06, | |
| "loss": 0.1416, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12376489490270615, | |
| "step": 3425, | |
| "valid_targets_mean": 3497.6, | |
| "valid_targets_min": 1274 | |
| }, | |
| { | |
| "epoch": 5.470494417862839, | |
| "grad_norm": 0.6832162024266292, | |
| "learning_rate": 5.55195690871211e-06, | |
| "loss": 0.1611, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19239741563796997, | |
| "step": 3430, | |
| "valid_targets_mean": 2752.0, | |
| "valid_targets_min": 1137 | |
| }, | |
| { | |
| "epoch": 5.478468899521531, | |
| "grad_norm": 0.6673288640941466, | |
| "learning_rate": 5.497075675178727e-06, | |
| "loss": 0.1628, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12479984760284424, | |
| "step": 3435, | |
| "valid_targets_mean": 2185.1, | |
| "valid_targets_min": 998 | |
| }, | |
| { | |
| "epoch": 5.486443381180223, | |
| "grad_norm": 0.6827660786954963, | |
| "learning_rate": 5.442423792610118e-06, | |
| "loss": 0.1624, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15979987382888794, | |
| "step": 3440, | |
| "valid_targets_mean": 2559.2, | |
| "valid_targets_min": 789 | |
| }, | |
| { | |
| "epoch": 5.494417862838915, | |
| "grad_norm": 0.6648142440724466, | |
| "learning_rate": 5.388002125277627e-06, | |
| "loss": 0.1548, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17778322100639343, | |
| "step": 3445, | |
| "valid_targets_mean": 2779.8, | |
| "valid_targets_min": 1059 | |
| }, | |
| { | |
| "epoch": 5.502392344497608, | |
| "grad_norm": 0.6355104542187777, | |
| "learning_rate": 5.333811533811945e-06, | |
| "loss": 0.1729, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2083851397037506, | |
| "step": 3450, | |
| "valid_targets_mean": 3374.8, | |
| "valid_targets_min": 1312 | |
| }, | |
| { | |
| "epoch": 5.5103668261563, | |
| "grad_norm": 0.6324432750904236, | |
| "learning_rate": 5.2798528751895265e-06, | |
| "loss": 0.1746, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14686718583106995, | |
| "step": 3455, | |
| "valid_targets_mean": 3143.6, | |
| "valid_targets_min": 530 | |
| }, | |
| { | |
| "epoch": 5.518341307814992, | |
| "grad_norm": 0.64304470738957, | |
| "learning_rate": 5.226127002718984e-06, | |
| "loss": 0.1391, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13552545011043549, | |
| "step": 3460, | |
| "valid_targets_mean": 2855.8, | |
| "valid_targets_min": 925 | |
| }, | |
| { | |
| "epoch": 5.526315789473684, | |
| "grad_norm": 0.6068488328932702, | |
| "learning_rate": 5.1726347660276424e-06, | |
| "loss": 0.1479, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1278594583272934, | |
| "step": 3465, | |
| "valid_targets_mean": 2640.1, | |
| "valid_targets_min": 1123 | |
| }, | |
| { | |
| "epoch": 5.5342902711323765, | |
| "grad_norm": 0.6186581531893287, | |
| "learning_rate": 5.119377011048066e-06, | |
| "loss": 0.1672, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18196257948875427, | |
| "step": 3470, | |
| "valid_targets_mean": 3513.9, | |
| "valid_targets_min": 1403 | |
| }, | |
| { | |
| "epoch": 5.542264752791069, | |
| "grad_norm": 0.6669804977769721, | |
| "learning_rate": 5.066354580004713e-06, | |
| "loss": 0.1584, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15941552817821503, | |
| "step": 3475, | |
| "valid_targets_mean": 2656.9, | |
| "valid_targets_min": 1576 | |
| }, | |
| { | |
| "epoch": 5.55023923444976, | |
| "grad_norm": 0.5584247466984443, | |
| "learning_rate": 5.013568311400599e-06, | |
| "loss": 0.1418, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1496984362602234, | |
| "step": 3480, | |
| "valid_targets_mean": 3440.2, | |
| "valid_targets_min": 1432 | |
| }, | |
| { | |
| "epoch": 5.558213716108453, | |
| "grad_norm": 0.5968686738437612, | |
| "learning_rate": 4.96101904000402e-06, | |
| "loss": 0.1619, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17291352152824402, | |
| "step": 3485, | |
| "valid_targets_mean": 3329.5, | |
| "valid_targets_min": 1615 | |
| }, | |
| { | |
| "epoch": 5.566188197767145, | |
| "grad_norm": 0.6237372043857617, | |
| "learning_rate": 4.908707596835396e-06, | |
| "loss": 0.1515, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12298108637332916, | |
| "step": 3490, | |
| "valid_targets_mean": 2981.9, | |
| "valid_targets_min": 1313 | |
| }, | |
| { | |
| "epoch": 5.574162679425838, | |
| "grad_norm": 0.6346515143599728, | |
| "learning_rate": 4.856634809154093e-06, | |
| "loss": 0.1654, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1985761821269989, | |
| "step": 3495, | |
| "valid_targets_mean": 3093.9, | |
| "valid_targets_min": 1159 | |
| }, | |
| { | |
| "epoch": 5.582137161084529, | |
| "grad_norm": 0.6081952670860399, | |
| "learning_rate": 4.804801500445338e-06, | |
| "loss": 0.1788, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17497466504573822, | |
| "step": 3500, | |
| "valid_targets_mean": 3463.1, | |
| "valid_targets_min": 471 | |
| }, | |
| { | |
| "epoch": 5.590111642743222, | |
| "grad_norm": 0.5677532822167262, | |
| "learning_rate": 4.753208490407233e-06, | |
| "loss": 0.1524, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1865505427122116, | |
| "step": 3505, | |
| "valid_targets_mean": 3796.1, | |
| "valid_targets_min": 1006 | |
| }, | |
| { | |
| "epoch": 5.598086124401914, | |
| "grad_norm": 0.6876441442315494, | |
| "learning_rate": 4.701856594937744e-06, | |
| "loss": 0.1466, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17622287571430206, | |
| "step": 3510, | |
| "valid_targets_mean": 2459.9, | |
| "valid_targets_min": 880 | |
| }, | |
| { | |
| "epoch": 5.606060606060606, | |
| "grad_norm": 0.6121241994834047, | |
| "learning_rate": 4.650746626121838e-06, | |
| "loss": 0.1863, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16413599252700806, | |
| "step": 3515, | |
| "valid_targets_mean": 2919.9, | |
| "valid_targets_min": 1113 | |
| }, | |
| { | |
| "epoch": 5.614035087719298, | |
| "grad_norm": 0.6187331081816392, | |
| "learning_rate": 4.5998793922186315e-06, | |
| "loss": 0.1413, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1643972545862198, | |
| "step": 3520, | |
| "valid_targets_mean": 3124.5, | |
| "valid_targets_min": 1606 | |
| }, | |
| { | |
| "epoch": 5.62200956937799, | |
| "grad_norm": 0.578937807626536, | |
| "learning_rate": 4.549255697648576e-06, | |
| "loss": 0.1387, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14879783987998962, | |
| "step": 3525, | |
| "valid_targets_mean": 2991.8, | |
| "valid_targets_min": 1059 | |
| }, | |
| { | |
| "epoch": 5.629984051036683, | |
| "grad_norm": 0.558863133501645, | |
| "learning_rate": 4.498876342980796e-06, | |
| "loss": 0.1311, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16602568328380585, | |
| "step": 3530, | |
| "valid_targets_mean": 3746.1, | |
| "valid_targets_min": 1425 | |
| }, | |
| { | |
| "epoch": 5.637958532695375, | |
| "grad_norm": 0.5968740877187344, | |
| "learning_rate": 4.448742124920368e-06, | |
| "loss": 0.1513, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16445884108543396, | |
| "step": 3535, | |
| "valid_targets_mean": 3706.6, | |
| "valid_targets_min": 1182 | |
| }, | |
| { | |
| "epoch": 5.645933014354067, | |
| "grad_norm": 0.59401001935693, | |
| "learning_rate": 4.39885383629576e-06, | |
| "loss": 0.1424, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12195871025323868, | |
| "step": 3540, | |
| "valid_targets_mean": 2801.0, | |
| "valid_targets_min": 1402 | |
| }, | |
| { | |
| "epoch": 5.653907496012759, | |
| "grad_norm": 0.6810295811672586, | |
| "learning_rate": 4.349212266046285e-06, | |
| "loss": 0.1475, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14159944653511047, | |
| "step": 3545, | |
| "valid_targets_mean": 2317.9, | |
| "valid_targets_min": 1367 | |
| }, | |
| { | |
| "epoch": 5.6618819776714515, | |
| "grad_norm": 0.5657955138742755, | |
| "learning_rate": 4.299818199209629e-06, | |
| "loss": 0.1456, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13119353353977203, | |
| "step": 3550, | |
| "valid_targets_mean": 3540.1, | |
| "valid_targets_min": 1045 | |
| }, | |
| { | |
| "epoch": 5.669856459330144, | |
| "grad_norm": 0.8104409273690198, | |
| "learning_rate": 4.250672416909407e-06, | |
| "loss": 0.1576, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1876513808965683, | |
| "step": 3555, | |
| "valid_targets_mean": 3162.6, | |
| "valid_targets_min": 1408 | |
| }, | |
| { | |
| "epoch": 5.6778309409888355, | |
| "grad_norm": 0.6124381093965978, | |
| "learning_rate": 4.201775696342862e-06, | |
| "loss": 0.1579, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13993270695209503, | |
| "step": 3560, | |
| "valid_targets_mean": 2776.8, | |
| "valid_targets_min": 1153 | |
| }, | |
| { | |
| "epoch": 5.685805422647528, | |
| "grad_norm": 0.6570663102638316, | |
| "learning_rate": 4.153128810768517e-06, | |
| "loss": 0.1452, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19519951939582825, | |
| "step": 3565, | |
| "valid_targets_mean": 3120.6, | |
| "valid_targets_min": 1299 | |
| }, | |
| { | |
| "epoch": 5.69377990430622, | |
| "grad_norm": 0.6605723879914376, | |
| "learning_rate": 4.104732529493991e-06, | |
| "loss": 0.1448, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13901135325431824, | |
| "step": 3570, | |
| "valid_targets_mean": 2307.8, | |
| "valid_targets_min": 1200 | |
| }, | |
| { | |
| "epoch": 5.701754385964913, | |
| "grad_norm": 0.6408169381973454, | |
| "learning_rate": 4.056587617863825e-06, | |
| "loss": 0.1483, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14372986555099487, | |
| "step": 3575, | |
| "valid_targets_mean": 2505.6, | |
| "valid_targets_min": 1173 | |
| }, | |
| { | |
| "epoch": 5.709728867623604, | |
| "grad_norm": 0.6526244361410768, | |
| "learning_rate": 4.008694837247345e-06, | |
| "loss": 0.1506, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15499386191368103, | |
| "step": 3580, | |
| "valid_targets_mean": 2597.3, | |
| "valid_targets_min": 1072 | |
| }, | |
| { | |
| "epoch": 5.717703349282297, | |
| "grad_norm": 0.7711661892385364, | |
| "learning_rate": 3.961054945026674e-06, | |
| "loss": 0.1375, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13299037516117096, | |
| "step": 3585, | |
| "valid_targets_mean": 2595.4, | |
| "valid_targets_min": 533 | |
| }, | |
| { | |
| "epoch": 5.725677830940989, | |
| "grad_norm": 0.5653622529411262, | |
| "learning_rate": 3.913668694584705e-06, | |
| "loss": 0.144, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1906440258026123, | |
| "step": 3590, | |
| "valid_targets_mean": 4048.9, | |
| "valid_targets_min": 1326 | |
| }, | |
| { | |
| "epoch": 5.733652312599681, | |
| "grad_norm": 0.6059994868500017, | |
| "learning_rate": 3.866536835293227e-06, | |
| "loss": 0.1645, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1835460066795349, | |
| "step": 3595, | |
| "valid_targets_mean": 3213.8, | |
| "valid_targets_min": 1082 | |
| }, | |
| { | |
| "epoch": 5.741626794258373, | |
| "grad_norm": 0.5632286313650832, | |
| "learning_rate": 3.819660112501053e-06, | |
| "loss": 0.1405, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11964814364910126, | |
| "step": 3600, | |
| "valid_targets_mean": 3211.3, | |
| "valid_targets_min": 1064 | |
| }, | |
| { | |
| "epoch": 5.749601275917065, | |
| "grad_norm": 0.5656484979331154, | |
| "learning_rate": 3.773039267522227e-06, | |
| "loss": 0.1411, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13798847794532776, | |
| "step": 3605, | |
| "valid_targets_mean": 3497.9, | |
| "valid_targets_min": 1097 | |
| }, | |
| { | |
| "epoch": 5.757575757575758, | |
| "grad_norm": 0.6669280692908005, | |
| "learning_rate": 3.72667503762433e-06, | |
| "loss": 0.1426, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12836626172065735, | |
| "step": 3610, | |
| "valid_targets_mean": 2467.6, | |
| "valid_targets_min": 812 | |
| }, | |
| { | |
| "epoch": 5.76555023923445, | |
| "grad_norm": 0.6105128759695477, | |
| "learning_rate": 3.680568156016786e-06, | |
| "loss": 0.1619, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21265044808387756, | |
| "step": 3615, | |
| "valid_targets_mean": 3298.8, | |
| "valid_targets_min": 1230 | |
| }, | |
| { | |
| "epoch": 5.773524720893142, | |
| "grad_norm": 0.7614562771653794, | |
| "learning_rate": 3.6347193518392776e-06, | |
| "loss": 0.1462, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14158612489700317, | |
| "step": 3620, | |
| "valid_targets_mean": 2508.9, | |
| "valid_targets_min": 862 | |
| }, | |
| { | |
| "epoch": 5.781499202551834, | |
| "grad_norm": 0.5871957486109551, | |
| "learning_rate": 3.58912935015024e-06, | |
| "loss": 0.1487, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1426844745874405, | |
| "step": 3625, | |
| "valid_targets_mean": 3297.4, | |
| "valid_targets_min": 1166 | |
| }, | |
| { | |
| "epoch": 5.7894736842105265, | |
| "grad_norm": 0.7064727598286994, | |
| "learning_rate": 3.543798871915367e-06, | |
| "loss": 0.1575, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.154343381524086, | |
| "step": 3630, | |
| "valid_targets_mean": 2664.8, | |
| "valid_targets_min": 876 | |
| }, | |
| { | |
| "epoch": 5.797448165869218, | |
| "grad_norm": 0.6633685511467207, | |
| "learning_rate": 3.498728633996209e-06, | |
| "loss": 0.1357, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12906265258789062, | |
| "step": 3635, | |
| "valid_targets_mean": 2260.7, | |
| "valid_targets_min": 1088 | |
| }, | |
| { | |
| "epoch": 5.8054226475279105, | |
| "grad_norm": 0.6214965202746205, | |
| "learning_rate": 3.453919349138859e-06, | |
| "loss": 0.1423, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11440666764974594, | |
| "step": 3640, | |
| "valid_targets_mean": 2508.8, | |
| "valid_targets_min": 1399 | |
| }, | |
| { | |
| "epoch": 5.813397129186603, | |
| "grad_norm": 0.5769112478791388, | |
| "learning_rate": 3.4093717259626514e-06, | |
| "loss": 0.1699, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11280746757984161, | |
| "step": 3645, | |
| "valid_targets_mean": 2709.9, | |
| "valid_targets_min": 1568 | |
| }, | |
| { | |
| "epoch": 5.821371610845295, | |
| "grad_norm": 0.6463824538730006, | |
| "learning_rate": 3.365086468948988e-06, | |
| "loss": 0.173, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1195705458521843, | |
| "step": 3650, | |
| "valid_targets_mean": 2292.4, | |
| "valid_targets_min": 1154 | |
| }, | |
| { | |
| "epoch": 5.829346092503988, | |
| "grad_norm": 0.6391150749806931, | |
| "learning_rate": 3.321064278430175e-06, | |
| "loss": 0.1611, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1568872481584549, | |
| "step": 3655, | |
| "valid_targets_mean": 3043.6, | |
| "valid_targets_min": 1213 | |
| }, | |
| { | |
| "epoch": 5.837320574162679, | |
| "grad_norm": 0.519107762034051, | |
| "learning_rate": 3.277305850578345e-06, | |
| "loss": 0.1359, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11462641507387161, | |
| "step": 3660, | |
| "valid_targets_mean": 3768.6, | |
| "valid_targets_min": 1037 | |
| }, | |
| { | |
| "epoch": 5.845295055821372, | |
| "grad_norm": 0.5558282842216485, | |
| "learning_rate": 3.2338118773944684e-06, | |
| "loss": 0.1562, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13219889998435974, | |
| "step": 3665, | |
| "valid_targets_mean": 3827.9, | |
| "valid_targets_min": 1308 | |
| }, | |
| { | |
| "epoch": 5.853269537480064, | |
| "grad_norm": 0.712652647722326, | |
| "learning_rate": 3.1905830466973975e-06, | |
| "loss": 0.1713, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2413114309310913, | |
| "step": 3670, | |
| "valid_targets_mean": 3104.1, | |
| "valid_targets_min": 1377 | |
| }, | |
| { | |
| "epoch": 5.861244019138756, | |
| "grad_norm": 0.8087033611981687, | |
| "learning_rate": 3.14762004211298e-06, | |
| "loss": 0.1499, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13118065893650055, | |
| "step": 3675, | |
| "valid_targets_mean": 2179.2, | |
| "valid_targets_min": 1057 | |
| }, | |
| { | |
| "epoch": 5.869218500797448, | |
| "grad_norm": 0.6965194625677074, | |
| "learning_rate": 3.1049235430632696e-06, | |
| "loss": 0.1648, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17841684818267822, | |
| "step": 3680, | |
| "valid_targets_mean": 2457.4, | |
| "valid_targets_min": 1164 | |
| }, | |
| { | |
| "epoch": 5.87719298245614, | |
| "grad_norm": 0.6429474542089771, | |
| "learning_rate": 3.062494224755759e-06, | |
| "loss": 0.1586, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.20610585808753967, | |
| "step": 3685, | |
| "valid_targets_mean": 3904.6, | |
| "valid_targets_min": 1201 | |
| }, | |
| { | |
| "epoch": 5.885167464114833, | |
| "grad_norm": 0.6122324755768935, | |
| "learning_rate": 3.0203327581727195e-06, | |
| "loss": 0.1431, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11550375819206238, | |
| "step": 3690, | |
| "valid_targets_mean": 2673.4, | |
| "valid_targets_min": 1456 | |
| }, | |
| { | |
| "epoch": 5.893141945773524, | |
| "grad_norm": 0.6890872162742891, | |
| "learning_rate": 2.9784398100605937e-06, | |
| "loss": 0.1447, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17845694720745087, | |
| "step": 3695, | |
| "valid_targets_mean": 2944.2, | |
| "valid_targets_min": 921 | |
| }, | |
| { | |
| "epoch": 5.901116427432217, | |
| "grad_norm": 0.6524913255544141, | |
| "learning_rate": 2.9368160429194127e-06, | |
| "loss": 0.1511, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1449909806251526, | |
| "step": 3700, | |
| "valid_targets_mean": 2448.1, | |
| "valid_targets_min": 1127 | |
| }, | |
| { | |
| "epoch": 5.909090909090909, | |
| "grad_norm": 0.5680543220274574, | |
| "learning_rate": 2.895462114992371e-06, | |
| "loss": 0.1521, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11523935198783875, | |
| "step": 3705, | |
| "valid_targets_mean": 2600.8, | |
| "valid_targets_min": 1253 | |
| }, | |
| { | |
| "epoch": 5.917065390749602, | |
| "grad_norm": 0.6588147712991279, | |
| "learning_rate": 2.8543786802553943e-06, | |
| "loss": 0.1468, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1229035034775734, | |
| "step": 3710, | |
| "valid_targets_mean": 2522.4, | |
| "valid_targets_min": 999 | |
| }, | |
| { | |
| "epoch": 5.925039872408293, | |
| "grad_norm": 0.7745513255839773, | |
| "learning_rate": 2.813566388406781e-06, | |
| "loss": 0.1586, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12802991271018982, | |
| "step": 3715, | |
| "valid_targets_mean": 2456.6, | |
| "valid_targets_min": 1135 | |
| }, | |
| { | |
| "epoch": 5.9330143540669855, | |
| "grad_norm": 0.7446954349027073, | |
| "learning_rate": 2.773025884856957e-06, | |
| "loss": 0.157, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17623551189899445, | |
| "step": 3720, | |
| "valid_targets_mean": 2696.7, | |
| "valid_targets_min": 1109 | |
| }, | |
| { | |
| "epoch": 5.940988835725678, | |
| "grad_norm": 0.6780845611384435, | |
| "learning_rate": 2.7327578107182585e-06, | |
| "loss": 0.1488, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13724076747894287, | |
| "step": 3725, | |
| "valid_targets_mean": 2162.1, | |
| "valid_targets_min": 1202 | |
| }, | |
| { | |
| "epoch": 5.94896331738437, | |
| "grad_norm": 0.6943638613747845, | |
| "learning_rate": 2.692762802794775e-06, | |
| "loss": 0.1502, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17806532979011536, | |
| "step": 3730, | |
| "valid_targets_mean": 2529.4, | |
| "valid_targets_min": 1031 | |
| }, | |
| { | |
| "epoch": 5.956937799043062, | |
| "grad_norm": 0.6260683420021232, | |
| "learning_rate": 2.6530414935723104e-06, | |
| "loss": 0.1483, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14727982878684998, | |
| "step": 3735, | |
| "valid_targets_mean": 3022.1, | |
| "valid_targets_min": 1694 | |
| }, | |
| { | |
| "epoch": 5.964912280701754, | |
| "grad_norm": 0.5711184251561199, | |
| "learning_rate": 2.6135945112083506e-06, | |
| "loss": 0.1753, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2091876119375229, | |
| "step": 3740, | |
| "valid_targets_mean": 3923.7, | |
| "valid_targets_min": 986 | |
| }, | |
| { | |
| "epoch": 5.972886762360447, | |
| "grad_norm": 0.660900776552596, | |
| "learning_rate": 2.574422479522156e-06, | |
| "loss": 0.1426, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14469899237155914, | |
| "step": 3745, | |
| "valid_targets_mean": 2478.1, | |
| "valid_targets_min": 1142 | |
| }, | |
| { | |
| "epoch": 5.980861244019139, | |
| "grad_norm": 0.6698862853262395, | |
| "learning_rate": 2.535526017984884e-06, | |
| "loss": 0.1586, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1301461160182953, | |
| "step": 3750, | |
| "valid_targets_mean": 2119.2, | |
| "valid_targets_min": 1058 | |
| }, | |
| { | |
| "epoch": 5.988835725677831, | |
| "grad_norm": 0.6050519500902528, | |
| "learning_rate": 2.4969057417097807e-06, | |
| "loss": 0.1389, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1766177862882614, | |
| "step": 3755, | |
| "valid_targets_mean": 3424.4, | |
| "valid_targets_min": 1367 | |
| }, | |
| { | |
| "epoch": 5.996810207336523, | |
| "grad_norm": 0.6388975080344989, | |
| "learning_rate": 2.458562261442483e-06, | |
| "loss": 0.1393, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12530069053173065, | |
| "step": 3760, | |
| "valid_targets_mean": 2602.0, | |
| "valid_targets_min": 1149 | |
| }, | |
| { | |
| "epoch": 6.0047846889952154, | |
| "grad_norm": 0.5447128366200804, | |
| "learning_rate": 2.4204961835513263e-06, | |
| "loss": 0.1464, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13561969995498657, | |
| "step": 3765, | |
| "valid_targets_mean": 3309.5, | |
| "valid_targets_min": 1247 | |
| }, | |
| { | |
| "epoch": 6.012759170653908, | |
| "grad_norm": 0.6139508350380686, | |
| "learning_rate": 2.3827081100177797e-06, | |
| "loss": 0.157, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.161412313580513, | |
| "step": 3770, | |
| "valid_targets_mean": 2966.3, | |
| "valid_targets_min": 1185 | |
| }, | |
| { | |
| "epoch": 6.020733652312599, | |
| "grad_norm": 0.6194921775961351, | |
| "learning_rate": 2.3451986384269266e-06, | |
| "loss": 0.1374, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1445777714252472, | |
| "step": 3775, | |
| "valid_targets_mean": 2521.9, | |
| "valid_targets_min": 1306 | |
| }, | |
| { | |
| "epoch": 6.028708133971292, | |
| "grad_norm": 0.637631777049995, | |
| "learning_rate": 2.307968361957993e-06, | |
| "loss": 0.1474, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2161223292350769, | |
| "step": 3780, | |
| "valid_targets_mean": 3306.6, | |
| "valid_targets_min": 1087 | |
| }, | |
| { | |
| "epoch": 6.036682615629984, | |
| "grad_norm": 0.738533022432783, | |
| "learning_rate": 2.2710178693749805e-06, | |
| "loss": 0.1557, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17869213223457336, | |
| "step": 3785, | |
| "valid_targets_mean": 2190.9, | |
| "valid_targets_min": 1312 | |
| }, | |
| { | |
| "epoch": 6.044657097288677, | |
| "grad_norm": 0.6527674057226728, | |
| "learning_rate": 2.2343477450173665e-06, | |
| "loss": 0.1469, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1218600869178772, | |
| "step": 3790, | |
| "valid_targets_mean": 2779.9, | |
| "valid_targets_min": 1470 | |
| }, | |
| { | |
| "epoch": 6.052631578947368, | |
| "grad_norm": 0.6630351920280422, | |
| "learning_rate": 2.197958568790839e-06, | |
| "loss": 0.1578, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17970821261405945, | |
| "step": 3795, | |
| "valid_targets_mean": 3350.1, | |
| "valid_targets_min": 982 | |
| }, | |
| { | |
| "epoch": 6.0606060606060606, | |
| "grad_norm": 0.5672542193779837, | |
| "learning_rate": 2.161850916158148e-06, | |
| "loss": 0.1267, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12792956829071045, | |
| "step": 3800, | |
| "valid_targets_mean": 3574.1, | |
| "valid_targets_min": 1535 | |
| }, | |
| { | |
| "epoch": 6.068580542264753, | |
| "grad_norm": 0.6123013687625073, | |
| "learning_rate": 2.1260253581299996e-06, | |
| "loss": 0.135, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13635091483592987, | |
| "step": 3805, | |
| "valid_targets_mean": 3412.9, | |
| "valid_targets_min": 1182 | |
| }, | |
| { | |
| "epoch": 6.076555023923445, | |
| "grad_norm": 0.6534334160731141, | |
| "learning_rate": 2.0904824612560046e-06, | |
| "loss": 0.1503, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.125827357172966, | |
| "step": 3810, | |
| "valid_targets_mean": 2346.2, | |
| "valid_targets_min": 1172 | |
| }, | |
| { | |
| "epoch": 6.084529505582137, | |
| "grad_norm": 0.6294651199199671, | |
| "learning_rate": 2.0552227876157536e-06, | |
| "loss": 0.1283, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12857142090797424, | |
| "step": 3815, | |
| "valid_targets_mean": 2880.9, | |
| "valid_targets_min": 1383 | |
| }, | |
| { | |
| "epoch": 6.092503987240829, | |
| "grad_norm": 0.6391181961066408, | |
| "learning_rate": 2.020246894809912e-06, | |
| "loss": 0.1473, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14293187856674194, | |
| "step": 3820, | |
| "valid_targets_mean": 2485.8, | |
| "valid_targets_min": 996 | |
| }, | |
| { | |
| "epoch": 6.100478468899522, | |
| "grad_norm": 0.7234527348754299, | |
| "learning_rate": 1.9855553359513836e-06, | |
| "loss": 0.1298, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1257050782442093, | |
| "step": 3825, | |
| "valid_targets_mean": 2228.8, | |
| "valid_targets_min": 1082 | |
| }, | |
| { | |
| "epoch": 6.108452950558214, | |
| "grad_norm": 0.6192919926151812, | |
| "learning_rate": 1.9511486596566054e-06, | |
| "loss": 0.1596, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10207498073577881, | |
| "step": 3830, | |
| "valid_targets_mean": 2159.6, | |
| "valid_targets_min": 1140 | |
| }, | |
| { | |
| "epoch": 6.116427432216906, | |
| "grad_norm": 0.5359741447900245, | |
| "learning_rate": 1.917027410036825e-06, | |
| "loss": 0.1393, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1523303985595703, | |
| "step": 3835, | |
| "valid_targets_mean": 4409.8, | |
| "valid_targets_min": 1701 | |
| }, | |
| { | |
| "epoch": 6.124401913875598, | |
| "grad_norm": 0.6692422456172075, | |
| "learning_rate": 1.8831921266895348e-06, | |
| "loss": 0.1326, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1260911226272583, | |
| "step": 3840, | |
| "valid_targets_mean": 2306.9, | |
| "valid_targets_min": 1306 | |
| }, | |
| { | |
| "epoch": 6.1323763955342905, | |
| "grad_norm": 0.6423890090470709, | |
| "learning_rate": 1.8496433446899197e-06, | |
| "loss": 0.1463, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13464131951332092, | |
| "step": 3845, | |
| "valid_targets_mean": 2437.8, | |
| "valid_targets_min": 1043 | |
| }, | |
| { | |
| "epoch": 6.140350877192983, | |
| "grad_norm": 0.6902212059072099, | |
| "learning_rate": 1.8163815945823881e-06, | |
| "loss": 0.1488, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1537630259990692, | |
| "step": 3850, | |
| "valid_targets_mean": 2586.1, | |
| "valid_targets_min": 1231 | |
| }, | |
| { | |
| "epoch": 6.148325358851674, | |
| "grad_norm": 0.5846067426158729, | |
| "learning_rate": 1.7834074023722082e-06, | |
| "loss": 0.1369, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1573687642812729, | |
| "step": 3855, | |
| "valid_targets_mean": 3312.4, | |
| "valid_targets_min": 1595 | |
| }, | |
| { | |
| "epoch": 6.156299840510367, | |
| "grad_norm": 0.6205363998371608, | |
| "learning_rate": 1.7507212895171632e-06, | |
| "loss": 0.1628, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17224471271038055, | |
| "step": 3860, | |
| "valid_targets_mean": 3192.4, | |
| "valid_targets_min": 1121 | |
| }, | |
| { | |
| "epoch": 6.164274322169059, | |
| "grad_norm": 0.6482621142553876, | |
| "learning_rate": 1.7183237729193081e-06, | |
| "loss": 0.1554, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14119577407836914, | |
| "step": 3865, | |
| "valid_targets_mean": 3200.1, | |
| "valid_targets_min": 987 | |
| }, | |
| { | |
| "epoch": 6.172248803827751, | |
| "grad_norm": 0.5945417763389266, | |
| "learning_rate": 1.6862153649168211e-06, | |
| "loss": 0.1667, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15197794139385223, | |
| "step": 3870, | |
| "valid_targets_mean": 4120.8, | |
| "valid_targets_min": 1689 | |
| }, | |
| { | |
| "epoch": 6.180223285486443, | |
| "grad_norm": 0.6424920862651418, | |
| "learning_rate": 1.6543965732758737e-06, | |
| "loss": 0.1434, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16543099284172058, | |
| "step": 3875, | |
| "valid_targets_mean": 3316.8, | |
| "valid_targets_min": 854 | |
| }, | |
| { | |
| "epoch": 6.188197767145136, | |
| "grad_norm": 0.7111768445227941, | |
| "learning_rate": 1.6228679011826032e-06, | |
| "loss": 0.1605, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13221696019172668, | |
| "step": 3880, | |
| "valid_targets_mean": 2675.8, | |
| "valid_targets_min": 1192 | |
| }, | |
| { | |
| "epoch": 6.196172248803828, | |
| "grad_norm": 0.5797517020648623, | |
| "learning_rate": 1.591629847235172e-06, | |
| "loss": 0.146, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.18317551910877228, | |
| "step": 3885, | |
| "valid_targets_mean": 3720.1, | |
| "valid_targets_min": 1166 | |
| }, | |
| { | |
| "epoch": 6.2041467304625195, | |
| "grad_norm": 0.6961211896005433, | |
| "learning_rate": 1.5606829054358686e-06, | |
| "loss": 0.1636, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1390981823205948, | |
| "step": 3890, | |
| "valid_targets_mean": 2520.2, | |
| "valid_targets_min": 789 | |
| }, | |
| { | |
| "epoch": 6.212121212121212, | |
| "grad_norm": 0.647792425598567, | |
| "learning_rate": 1.5300275651832963e-06, | |
| "loss": 0.1504, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1477465182542801, | |
| "step": 3895, | |
| "valid_targets_mean": 2948.1, | |
| "valid_targets_min": 1051 | |
| }, | |
| { | |
| "epoch": 6.220095693779904, | |
| "grad_norm": 0.6425614569659422, | |
| "learning_rate": 1.499664311264648e-06, | |
| "loss": 0.1331, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11879262328147888, | |
| "step": 3900, | |
| "valid_targets_mean": 2411.4, | |
| "valid_targets_min": 803 | |
| }, | |
| { | |
| "epoch": 6.228070175438597, | |
| "grad_norm": 0.6560346228896774, | |
| "learning_rate": 1.4695936238480135e-06, | |
| "loss": 0.1544, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11653897911310196, | |
| "step": 3905, | |
| "valid_targets_mean": 2464.3, | |
| "valid_targets_min": 1270 | |
| }, | |
| { | |
| "epoch": 6.236044657097288, | |
| "grad_norm": 0.5913785898356396, | |
| "learning_rate": 1.4398159784748144e-06, | |
| "loss": 0.12, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11476568132638931, | |
| "step": 3910, | |
| "valid_targets_mean": 2866.1, | |
| "valid_targets_min": 1170 | |
| }, | |
| { | |
| "epoch": 6.244019138755981, | |
| "grad_norm": 0.6344887654680683, | |
| "learning_rate": 1.4103318460522598e-06, | |
| "loss": 0.1455, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11846456676721573, | |
| "step": 3915, | |
| "valid_targets_mean": 2540.3, | |
| "valid_targets_min": 947 | |
| }, | |
| { | |
| "epoch": 6.251993620414673, | |
| "grad_norm": 0.6352575814931524, | |
| "learning_rate": 1.3811416928459177e-06, | |
| "loss": 0.1463, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12151379138231277, | |
| "step": 3920, | |
| "valid_targets_mean": 3051.1, | |
| "valid_targets_min": 1049 | |
| }, | |
| { | |
| "epoch": 6.2599681020733655, | |
| "grad_norm": 0.6533137247391657, | |
| "learning_rate": 1.3522459804723353e-06, | |
| "loss": 0.1484, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15973787009716034, | |
| "step": 3925, | |
| "valid_targets_mean": 2525.4, | |
| "valid_targets_min": 1135 | |
| }, | |
| { | |
| "epoch": 6.267942583732057, | |
| "grad_norm": 0.6295268930003071, | |
| "learning_rate": 1.3236451658917293e-06, | |
| "loss": 0.1299, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1486193835735321, | |
| "step": 3930, | |
| "valid_targets_mean": 3116.2, | |
| "valid_targets_min": 1571 | |
| }, | |
| { | |
| "epoch": 6.2759170653907494, | |
| "grad_norm": 0.6830753046565643, | |
| "learning_rate": 1.2953397014007728e-06, | |
| "loss": 0.1393, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14981475472450256, | |
| "step": 3935, | |
| "valid_targets_mean": 2623.8, | |
| "valid_targets_min": 1333 | |
| }, | |
| { | |
| "epoch": 6.283891547049442, | |
| "grad_norm": 0.6028379965845788, | |
| "learning_rate": 1.2673300346254447e-06, | |
| "loss": 0.1402, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14106810092926025, | |
| "step": 3940, | |
| "valid_targets_mean": 3649.6, | |
| "valid_targets_min": 1156 | |
| }, | |
| { | |
| "epoch": 6.291866028708134, | |
| "grad_norm": 0.6231653564515409, | |
| "learning_rate": 1.239616608513925e-06, | |
| "loss": 0.1443, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.132007896900177, | |
| "step": 3945, | |
| "valid_targets_mean": 2914.5, | |
| "valid_targets_min": 1202 | |
| }, | |
| { | |
| "epoch": 6.299840510366826, | |
| "grad_norm": 0.7144670706226991, | |
| "learning_rate": 1.2121998613296259e-06, | |
| "loss": 0.1372, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1385582983493805, | |
| "step": 3950, | |
| "valid_targets_mean": 3412.4, | |
| "valid_targets_min": 1244 | |
| }, | |
| { | |
| "epoch": 6.307814992025518, | |
| "grad_norm": 0.6711423788649739, | |
| "learning_rate": 1.1850802266442396e-06, | |
| "loss": 0.1428, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1569822132587433, | |
| "step": 3955, | |
| "valid_targets_mean": 2483.0, | |
| "valid_targets_min": 1022 | |
| }, | |
| { | |
| "epoch": 6.315789473684211, | |
| "grad_norm": 0.6347098632735023, | |
| "learning_rate": 1.1582581333308784e-06, | |
| "loss": 0.1583, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17010065913200378, | |
| "step": 3960, | |
| "valid_targets_mean": 3410.2, | |
| "valid_targets_min": 1383 | |
| }, | |
| { | |
| "epoch": 6.323763955342903, | |
| "grad_norm": 0.7134750123487895, | |
| "learning_rate": 1.1317340055573122e-06, | |
| "loss": 0.1391, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14928308129310608, | |
| "step": 3965, | |
| "valid_targets_mean": 2386.5, | |
| "valid_targets_min": 533 | |
| }, | |
| { | |
| "epoch": 6.3317384370015946, | |
| "grad_norm": 0.7403552637539116, | |
| "learning_rate": 1.1055082627792357e-06, | |
| "loss": 0.1535, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1645737886428833, | |
| "step": 3970, | |
| "valid_targets_mean": 2855.1, | |
| "valid_targets_min": 1410 | |
| }, | |
| { | |
| "epoch": 6.339712918660287, | |
| "grad_norm": 0.7439603296881704, | |
| "learning_rate": 1.0795813197336602e-06, | |
| "loss": 0.1661, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16814234852790833, | |
| "step": 3975, | |
| "valid_targets_mean": 2641.1, | |
| "valid_targets_min": 1119 | |
| }, | |
| { | |
| "epoch": 6.347687400318979, | |
| "grad_norm": 0.8558358007646358, | |
| "learning_rate": 1.0539535864323391e-06, | |
| "loss": 0.1577, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17330396175384521, | |
| "step": 3980, | |
| "valid_targets_mean": 2434.4, | |
| "valid_targets_min": 1006 | |
| }, | |
| { | |
| "epoch": 6.355661881977672, | |
| "grad_norm": 0.596512645985537, | |
| "learning_rate": 1.0286254681552777e-06, | |
| "loss": 0.1474, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15838459134101868, | |
| "step": 3985, | |
| "valid_targets_mean": 3372.1, | |
| "valid_targets_min": 684 | |
| }, | |
| { | |
| "epoch": 6.363636363636363, | |
| "grad_norm": 0.6467431933291012, | |
| "learning_rate": 1.0035973654443466e-06, | |
| "loss": 0.1425, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12675225734710693, | |
| "step": 3990, | |
| "valid_targets_mean": 2303.3, | |
| "valid_targets_min": 1219 | |
| }, | |
| { | |
| "epoch": 6.371610845295056, | |
| "grad_norm": 0.6266060938418669, | |
| "learning_rate": 9.788696740969295e-07, | |
| "loss": 0.1675, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.19282695651054382, | |
| "step": 3995, | |
| "valid_targets_mean": 3250.9, | |
| "valid_targets_min": 1133 | |
| }, | |
| { | |
| "epoch": 6.379585326953748, | |
| "grad_norm": 0.603105276214445, | |
| "learning_rate": 9.544427851596661e-07, | |
| "loss": 0.1489, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1304808259010315, | |
| "step": 4000, | |
| "valid_targets_mean": 2783.1, | |
| "valid_targets_min": 858 | |
| }, | |
| { | |
| "epoch": 6.3875598086124405, | |
| "grad_norm": 0.6045810868867998, | |
| "learning_rate": 9.303170849222764e-07, | |
| "loss": 0.1457, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13092729449272156, | |
| "step": 4005, | |
| "valid_targets_mean": 3034.9, | |
| "valid_targets_min": 1082 | |
| }, | |
| { | |
| "epoch": 6.395534290271132, | |
| "grad_norm": 0.6725605538366902, | |
| "learning_rate": 9.064929549114421e-07, | |
| "loss": 0.1489, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13961663842201233, | |
| "step": 4010, | |
| "valid_targets_mean": 3021.6, | |
| "valid_targets_min": 1365 | |
| }, | |
| { | |
| "epoch": 6.4035087719298245, | |
| "grad_norm": 0.5280456680065018, | |
| "learning_rate": 8.829707718847835e-07, | |
| "loss": 0.1628, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21119526028633118, | |
| "step": 4015, | |
| "valid_targets_mean": 5034.1, | |
| "valid_targets_min": 1383 | |
| }, | |
| { | |
| "epoch": 6.411483253588517, | |
| "grad_norm": 0.6870027710411994, | |
| "learning_rate": 8.597509078248923e-07, | |
| "loss": 0.1373, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11793472617864609, | |
| "step": 4020, | |
| "valid_targets_mean": 3262.4, | |
| "valid_targets_min": 1268 | |
| }, | |
| { | |
| "epoch": 6.419457735247209, | |
| "grad_norm": 0.6636142716776369, | |
| "learning_rate": 8.368337299334461e-07, | |
| "loss": 0.1498, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15093785524368286, | |
| "step": 4025, | |
| "valid_targets_mean": 2730.2, | |
| "valid_targets_min": 1267 | |
| }, | |
| { | |
| "epoch": 6.427432216905901, | |
| "grad_norm": 0.6510224950909901, | |
| "learning_rate": 8.142196006254144e-07, | |
| "loss": 0.1526, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12001948803663254, | |
| "step": 4030, | |
| "valid_targets_mean": 2182.4, | |
| "valid_targets_min": 1179 | |
| }, | |
| { | |
| "epoch": 6.435406698564593, | |
| "grad_norm": 0.6245752131587334, | |
| "learning_rate": 7.919088775233264e-07, | |
| "loss": 0.1422, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11676537990570068, | |
| "step": 4035, | |
| "valid_targets_mean": 2794.9, | |
| "valid_targets_min": 1104 | |
| }, | |
| { | |
| "epoch": 6.443381180223286, | |
| "grad_norm": 0.7044813681400015, | |
| "learning_rate": 7.699019134515917e-07, | |
| "loss": 0.1469, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16797736287117004, | |
| "step": 4040, | |
| "valid_targets_mean": 2557.8, | |
| "valid_targets_min": 991 | |
| }, | |
| { | |
| "epoch": 6.451355661881978, | |
| "grad_norm": 0.6616002971033802, | |
| "learning_rate": 7.48199056430956e-07, | |
| "loss": 0.1454, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13709191977977753, | |
| "step": 4045, | |
| "valid_targets_mean": 2547.1, | |
| "valid_targets_min": 1067 | |
| }, | |
| { | |
| "epoch": 6.45933014354067, | |
| "grad_norm": 0.5983415327849058, | |
| "learning_rate": 7.268006496729762e-07, | |
| "loss": 0.1413, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1570061445236206, | |
| "step": 4050, | |
| "valid_targets_mean": 3875.0, | |
| "valid_targets_min": 1334 | |
| }, | |
| { | |
| "epoch": 6.467304625199362, | |
| "grad_norm": 0.5883938864827607, | |
| "learning_rate": 7.057070315745851e-07, | |
| "loss": 0.1454, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14021623134613037, | |
| "step": 4055, | |
| "valid_targets_mean": 3387.9, | |
| "valid_targets_min": 1158 | |
| }, | |
| { | |
| "epoch": 6.475279106858054, | |
| "grad_norm": 0.6381148460727919, | |
| "learning_rate": 6.849185357127686e-07, | |
| "loss": 0.1436, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1419590264558792, | |
| "step": 4060, | |
| "valid_targets_mean": 2658.4, | |
| "valid_targets_min": 1063 | |
| }, | |
| { | |
| "epoch": 6.483253588516747, | |
| "grad_norm": 0.5626153764312638, | |
| "learning_rate": 6.64435490839257e-07, | |
| "loss": 0.1882, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1861887127161026, | |
| "step": 4065, | |
| "valid_targets_mean": 4139.4, | |
| "valid_targets_min": 1263 | |
| }, | |
| { | |
| "epoch": 6.491228070175438, | |
| "grad_norm": 0.6357101886196256, | |
| "learning_rate": 6.442582208753578e-07, | |
| "loss": 0.1437, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15499606728553772, | |
| "step": 4070, | |
| "valid_targets_mean": 2689.3, | |
| "valid_targets_min": 1081 | |
| }, | |
| { | |
| "epoch": 6.499202551834131, | |
| "grad_norm": 0.6910304026583559, | |
| "learning_rate": 6.243870449068068e-07, | |
| "loss": 0.1741, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2073277235031128, | |
| "step": 4075, | |
| "valid_targets_mean": 2765.3, | |
| "valid_targets_min": 1104 | |
| }, | |
| { | |
| "epoch": 6.507177033492823, | |
| "grad_norm": 0.58851977473546, | |
| "learning_rate": 6.048222771787382e-07, | |
| "loss": 0.1526, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14743703603744507, | |
| "step": 4080, | |
| "valid_targets_mean": 3211.2, | |
| "valid_targets_min": 1356 | |
| }, | |
| { | |
| "epoch": 6.515151515151516, | |
| "grad_norm": 0.6178750502243548, | |
| "learning_rate": 5.85564227090707e-07, | |
| "loss": 0.1472, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17087917029857635, | |
| "step": 4085, | |
| "valid_targets_mean": 3659.9, | |
| "valid_targets_min": 1231 | |
| }, | |
| { | |
| "epoch": 6.523125996810207, | |
| "grad_norm": 0.6938061545052304, | |
| "learning_rate": 5.666131991917989e-07, | |
| "loss": 0.1333, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1300240010023117, | |
| "step": 4090, | |
| "valid_targets_mean": 2109.9, | |
| "valid_targets_min": 1371 | |
| }, | |
| { | |
| "epoch": 6.5311004784688995, | |
| "grad_norm": 0.6642575349820269, | |
| "learning_rate": 5.479694931758194e-07, | |
| "loss": 0.1434, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1611027717590332, | |
| "step": 4095, | |
| "valid_targets_mean": 2830.8, | |
| "valid_targets_min": 1066 | |
| }, | |
| { | |
| "epoch": 6.539074960127592, | |
| "grad_norm": 0.6070107059437827, | |
| "learning_rate": 5.296334038765483e-07, | |
| "loss": 0.1428, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13571247458457947, | |
| "step": 4100, | |
| "valid_targets_mean": 2914.1, | |
| "valid_targets_min": 1169 | |
| }, | |
| { | |
| "epoch": 6.5470494417862835, | |
| "grad_norm": 0.6293749348387128, | |
| "learning_rate": 5.116052212630696e-07, | |
| "loss": 0.1219, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12869954109191895, | |
| "step": 4105, | |
| "valid_targets_mean": 2649.8, | |
| "valid_targets_min": 1237 | |
| }, | |
| { | |
| "epoch": 6.555023923444976, | |
| "grad_norm": 0.6390191087235467, | |
| "learning_rate": 4.938852304352026e-07, | |
| "loss": 0.1565, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.2021748423576355, | |
| "step": 4110, | |
| "valid_targets_mean": 3688.9, | |
| "valid_targets_min": 961 | |
| }, | |
| { | |
| "epoch": 6.562998405103668, | |
| "grad_norm": 0.6627492608631983, | |
| "learning_rate": 4.7647371161898547e-07, | |
| "loss": 0.123, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11246272921562195, | |
| "step": 4115, | |
| "valid_targets_mean": 2175.8, | |
| "valid_targets_min": 1234 | |
| }, | |
| { | |
| "epoch": 6.570972886762361, | |
| "grad_norm": 0.62234154665454, | |
| "learning_rate": 4.593709401622359e-07, | |
| "loss": 0.1683, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13496442139148712, | |
| "step": 4120, | |
| "valid_targets_mean": 2949.5, | |
| "valid_targets_min": 1421 | |
| }, | |
| { | |
| "epoch": 6.578947368421053, | |
| "grad_norm": 0.605016951584934, | |
| "learning_rate": 4.425771865302153e-07, | |
| "loss": 0.1448, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14617837965488434, | |
| "step": 4125, | |
| "valid_targets_mean": 3346.9, | |
| "valid_targets_min": 1320 | |
| }, | |
| { | |
| "epoch": 6.586921850079745, | |
| "grad_norm": 0.7083006189079447, | |
| "learning_rate": 4.2609271630133174e-07, | |
| "loss": 0.1732, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17233526706695557, | |
| "step": 4130, | |
| "valid_targets_mean": 2765.6, | |
| "valid_targets_min": 1235 | |
| }, | |
| { | |
| "epoch": 6.594896331738437, | |
| "grad_norm": 0.6422713513507917, | |
| "learning_rate": 4.099177901629525e-07, | |
| "loss": 0.1546, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14676770567893982, | |
| "step": 4135, | |
| "valid_targets_mean": 2894.6, | |
| "valid_targets_min": 1098 | |
| }, | |
| { | |
| "epoch": 6.6028708133971294, | |
| "grad_norm": 0.5862133305303223, | |
| "learning_rate": 3.9405266390727836e-07, | |
| "loss": 0.143, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1555536389350891, | |
| "step": 4140, | |
| "valid_targets_mean": 4025.8, | |
| "valid_targets_min": 1297 | |
| }, | |
| { | |
| "epoch": 6.610845295055821, | |
| "grad_norm": 0.649117157563772, | |
| "learning_rate": 3.7849758842729344e-07, | |
| "loss": 0.1335, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13616587221622467, | |
| "step": 4145, | |
| "valid_targets_mean": 2954.0, | |
| "valid_targets_min": 1108 | |
| }, | |
| { | |
| "epoch": 6.618819776714513, | |
| "grad_norm": 0.7199166365178159, | |
| "learning_rate": 3.632528097128085e-07, | |
| "loss": 0.1491, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14053437113761902, | |
| "step": 4150, | |
| "valid_targets_mean": 2117.9, | |
| "valid_targets_min": 1069 | |
| }, | |
| { | |
| "epoch": 6.626794258373206, | |
| "grad_norm": 0.6425703556421833, | |
| "learning_rate": 3.48318568846564e-07, | |
| "loss": 0.1454, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14537614583969116, | |
| "step": 4155, | |
| "valid_targets_mean": 3269.3, | |
| "valid_targets_min": 1160 | |
| }, | |
| { | |
| "epoch": 6.634768740031898, | |
| "grad_norm": 0.6587464590369769, | |
| "learning_rate": 3.336951020004087e-07, | |
| "loss": 0.1276, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13485220074653625, | |
| "step": 4160, | |
| "valid_targets_mean": 2578.7, | |
| "valid_targets_min": 1018 | |
| }, | |
| { | |
| "epoch": 6.64274322169059, | |
| "grad_norm": 0.6768113096854557, | |
| "learning_rate": 3.1938264043158694e-07, | |
| "loss": 0.1319, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13307702541351318, | |
| "step": 4165, | |
| "valid_targets_mean": 2578.9, | |
| "valid_targets_min": 1201 | |
| }, | |
| { | |
| "epoch": 6.650717703349282, | |
| "grad_norm": 0.5890393534004379, | |
| "learning_rate": 3.05381410479062e-07, | |
| "loss": 0.1327, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11060281097888947, | |
| "step": 4170, | |
| "valid_targets_mean": 2939.9, | |
| "valid_targets_min": 1043 | |
| }, | |
| { | |
| "epoch": 6.6586921850079746, | |
| "grad_norm": 0.6522407226289458, | |
| "learning_rate": 2.916916335599407e-07, | |
| "loss": 0.1301, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13013817369937897, | |
| "step": 4175, | |
| "valid_targets_mean": 2759.8, | |
| "valid_targets_min": 1309 | |
| }, | |
| { | |
| "epoch": 6.666666666666667, | |
| "grad_norm": 0.6332293594628468, | |
| "learning_rate": 2.783135261659831e-07, | |
| "loss": 0.1254, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10817860811948776, | |
| "step": 4180, | |
| "valid_targets_mean": 2399.8, | |
| "valid_targets_min": 978 | |
| }, | |
| { | |
| "epoch": 6.6746411483253585, | |
| "grad_norm": 0.592325682164998, | |
| "learning_rate": 2.6524729986016293e-07, | |
| "loss": 0.1613, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13545538485050201, | |
| "step": 4185, | |
| "valid_targets_mean": 3311.6, | |
| "valid_targets_min": 1127 | |
| }, | |
| { | |
| "epoch": 6.682615629984051, | |
| "grad_norm": 0.673758679985898, | |
| "learning_rate": 2.524931612733328e-07, | |
| "loss": 0.1335, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13675081729888916, | |
| "step": 4190, | |
| "valid_targets_mean": 2591.3, | |
| "valid_targets_min": 1481 | |
| }, | |
| { | |
| "epoch": 6.690590111642743, | |
| "grad_norm": 0.6966576529900051, | |
| "learning_rate": 2.400513121009529e-07, | |
| "loss": 0.1458, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1509852409362793, | |
| "step": 4195, | |
| "valid_targets_mean": 2215.1, | |
| "valid_targets_min": 1370 | |
| }, | |
| { | |
| "epoch": 6.698564593301436, | |
| "grad_norm": 0.6478122100991643, | |
| "learning_rate": 2.279219490998985e-07, | |
| "loss": 0.1378, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1596289575099945, | |
| "step": 4200, | |
| "valid_targets_mean": 2952.1, | |
| "valid_targets_min": 1239 | |
| }, | |
| { | |
| "epoch": 6.706539074960127, | |
| "grad_norm": 0.6952489954446537, | |
| "learning_rate": 2.161052640853578e-07, | |
| "loss": 0.1526, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15504632890224457, | |
| "step": 4205, | |
| "valid_targets_mean": 3227.2, | |
| "valid_targets_min": 1237 | |
| }, | |
| { | |
| "epoch": 6.71451355661882, | |
| "grad_norm": 0.6514941213275741, | |
| "learning_rate": 2.0460144392778768e-07, | |
| "loss": 0.1444, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10687048733234406, | |
| "step": 4210, | |
| "valid_targets_mean": 2079.6, | |
| "valid_targets_min": 1159 | |
| }, | |
| { | |
| "epoch": 6.722488038277512, | |
| "grad_norm": 0.6063876513438099, | |
| "learning_rate": 1.9341067054996277e-07, | |
| "loss": 0.1388, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13883773982524872, | |
| "step": 4215, | |
| "valid_targets_mean": 2910.4, | |
| "valid_targets_min": 1058 | |
| }, | |
| { | |
| "epoch": 6.7304625199362045, | |
| "grad_norm": 0.6791687773478821, | |
| "learning_rate": 1.8253312092409992e-07, | |
| "loss": 0.1434, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15515685081481934, | |
| "step": 4220, | |
| "valid_targets_mean": 3073.0, | |
| "valid_targets_min": 1273 | |
| }, | |
| { | |
| "epoch": 6.738437001594896, | |
| "grad_norm": 0.6218041337473685, | |
| "learning_rate": 1.7196896706906273e-07, | |
| "loss": 0.1353, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11648586392402649, | |
| "step": 4225, | |
| "valid_targets_mean": 2936.0, | |
| "valid_targets_min": 965 | |
| }, | |
| { | |
| "epoch": 6.746411483253588, | |
| "grad_norm": 0.6617392762256765, | |
| "learning_rate": 1.6171837604762597e-07, | |
| "loss": 0.138, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16125863790512085, | |
| "step": 4230, | |
| "valid_targets_mean": 3518.8, | |
| "valid_targets_min": 1593 | |
| }, | |
| { | |
| "epoch": 6.754385964912281, | |
| "grad_norm": 0.6784609926009033, | |
| "learning_rate": 1.5178150996385755e-07, | |
| "loss": 0.1372, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12986430525779724, | |
| "step": 4235, | |
| "valid_targets_mean": 2628.8, | |
| "valid_targets_min": 1094 | |
| }, | |
| { | |
| "epoch": 6.762360446570973, | |
| "grad_norm": 0.696201617201948, | |
| "learning_rate": 1.421585259605318e-07, | |
| "loss": 0.128, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11709525436162949, | |
| "step": 4240, | |
| "valid_targets_mean": 1826.1, | |
| "valid_targets_min": 950 | |
| }, | |
| { | |
| "epoch": 6.770334928229665, | |
| "grad_norm": 0.6367805811916879, | |
| "learning_rate": 1.3284957621666039e-07, | |
| "loss": 0.1326, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.11604160070419312, | |
| "step": 4245, | |
| "valid_targets_mean": 2452.1, | |
| "valid_targets_min": 1352 | |
| }, | |
| { | |
| "epoch": 6.778309409888357, | |
| "grad_norm": 0.6346156166777578, | |
| "learning_rate": 1.2385480794507853e-07, | |
| "loss": 0.1226, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1615995466709137, | |
| "step": 4250, | |
| "valid_targets_mean": 2704.9, | |
| "valid_targets_min": 1101 | |
| }, | |
| { | |
| "epoch": 6.78628389154705, | |
| "grad_norm": 0.6614819884519686, | |
| "learning_rate": 1.1517436339011589e-07, | |
| "loss": 0.1269, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1173841580748558, | |
| "step": 4255, | |
| "valid_targets_mean": 2531.2, | |
| "valid_targets_min": 1025 | |
| }, | |
| { | |
| "epoch": 6.794258373205742, | |
| "grad_norm": 0.5967106394852402, | |
| "learning_rate": 1.0680837982535607e-07, | |
| "loss": 0.1448, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10654541105031967, | |
| "step": 4260, | |
| "valid_targets_mean": 2876.8, | |
| "valid_targets_min": 1046 | |
| }, | |
| { | |
| "epoch": 6.8022328548644335, | |
| "grad_norm": 0.6070899574042022, | |
| "learning_rate": 9.875698955145174e-08, | |
| "loss": 0.1476, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14284199476242065, | |
| "step": 4265, | |
| "valid_targets_mean": 3178.1, | |
| "valid_targets_min": 1513 | |
| }, | |
| { | |
| "epoch": 6.810207336523126, | |
| "grad_norm": 0.6780997054136256, | |
| "learning_rate": 9.102031989404403e-08, | |
| "loss": 0.141, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16974301636219025, | |
| "step": 4270, | |
| "valid_targets_mean": 2744.4, | |
| "valid_targets_min": 997 | |
| }, | |
| { | |
| "epoch": 6.818181818181818, | |
| "grad_norm": 0.6754612475098248, | |
| "learning_rate": 8.359849320174196e-08, | |
| "loss": 0.1384, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14120808243751526, | |
| "step": 4275, | |
| "valid_targets_mean": 2615.7, | |
| "valid_targets_min": 1361 | |
| }, | |
| { | |
| "epoch": 6.826156299840511, | |
| "grad_norm": 0.693794580517889, | |
| "learning_rate": 7.649162684419731e-08, | |
| "loss": 0.1353, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12366122752428055, | |
| "step": 4280, | |
| "valid_targets_mean": 2165.4, | |
| "valid_targets_min": 930 | |
| }, | |
| { | |
| "epoch": 6.834130781499202, | |
| "grad_norm": 0.6818500018106061, | |
| "learning_rate": 6.969983321023499e-08, | |
| "loss": 0.1479, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14141970872879028, | |
| "step": 4285, | |
| "valid_targets_mean": 2466.6, | |
| "valid_targets_min": 1200 | |
| }, | |
| { | |
| "epoch": 6.842105263157895, | |
| "grad_norm": 0.5987656732671883, | |
| "learning_rate": 6.322321970608337e-08, | |
| "loss": 0.1343, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16968190670013428, | |
| "step": 4290, | |
| "valid_targets_mean": 3152.5, | |
| "valid_targets_min": 1227 | |
| }, | |
| { | |
| "epoch": 6.850079744816587, | |
| "grad_norm": 0.6439966316378877, | |
| "learning_rate": 5.7061888753677796e-08, | |
| "loss": 0.1444, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14066863059997559, | |
| "step": 4295, | |
| "valid_targets_mean": 2770.7, | |
| "valid_targets_min": 1023 | |
| }, | |
| { | |
| "epoch": 6.858054226475279, | |
| "grad_norm": 0.6544479541120145, | |
| "learning_rate": 5.121593778903755e-08, | |
| "loss": 0.1589, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.17600896954536438, | |
| "step": 4300, | |
| "valid_targets_mean": 3378.7, | |
| "valid_targets_min": 1227 | |
| }, | |
| { | |
| "epoch": 6.866028708133971, | |
| "grad_norm": 0.5672287636141157, | |
| "learning_rate": 4.5685459260722544e-08, | |
| "loss": 0.1375, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21883684396743774, | |
| "step": 4305, | |
| "valid_targets_mean": 4971.3, | |
| "valid_targets_min": 975 | |
| }, | |
| { | |
| "epoch": 6.8740031897926634, | |
| "grad_norm": 0.6252371250774587, | |
| "learning_rate": 4.047054062837452e-08, | |
| "loss": 0.1559, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13992947340011597, | |
| "step": 4310, | |
| "valid_targets_mean": 2978.4, | |
| "valid_targets_min": 990 | |
| }, | |
| { | |
| "epoch": 6.881977671451356, | |
| "grad_norm": 0.6625804873828336, | |
| "learning_rate": 3.55712643613404e-08, | |
| "loss": 0.1388, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.14451831579208374, | |
| "step": 4315, | |
| "valid_targets_mean": 2907.2, | |
| "valid_targets_min": 1053 | |
| }, | |
| { | |
| "epoch": 6.889952153110048, | |
| "grad_norm": 0.9522650055323967, | |
| "learning_rate": 3.0987707937351066e-08, | |
| "loss": 0.181, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.21254639327526093, | |
| "step": 4320, | |
| "valid_targets_mean": 2916.5, | |
| "valid_targets_min": 1095 | |
| }, | |
| { | |
| "epoch": 6.89792663476874, | |
| "grad_norm": 0.7728711461821048, | |
| "learning_rate": 2.6719943841311268e-08, | |
| "loss": 0.1444, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15055662393569946, | |
| "step": 4325, | |
| "valid_targets_mean": 2948.6, | |
| "valid_targets_min": 1408 | |
| }, | |
| { | |
| "epoch": 6.905901116427432, | |
| "grad_norm": 0.6584848163843601, | |
| "learning_rate": 2.2768039564151635e-08, | |
| "loss": 0.1317, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13766586780548096, | |
| "step": 4330, | |
| "valid_targets_mean": 2726.1, | |
| "valid_targets_min": 1372 | |
| }, | |
| { | |
| "epoch": 6.913875598086125, | |
| "grad_norm": 0.6431909009393747, | |
| "learning_rate": 1.913205760175174e-08, | |
| "loss": 0.146, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15910640358924866, | |
| "step": 4335, | |
| "valid_targets_mean": 3045.4, | |
| "valid_targets_min": 1098 | |
| }, | |
| { | |
| "epoch": 6.921850079744816, | |
| "grad_norm": 0.5604331917524619, | |
| "learning_rate": 1.5812055453963136e-08, | |
| "loss": 0.123, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1279834806919098, | |
| "step": 4340, | |
| "valid_targets_mean": 3657.7, | |
| "valid_targets_min": 1548 | |
| }, | |
| { | |
| "epoch": 6.9298245614035086, | |
| "grad_norm": 0.5809407923917316, | |
| "learning_rate": 1.280808562369229e-08, | |
| "loss": 0.1573, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1330064982175827, | |
| "step": 4345, | |
| "valid_targets_mean": 3600.9, | |
| "valid_targets_min": 1044 | |
| }, | |
| { | |
| "epoch": 6.937799043062201, | |
| "grad_norm": 0.621224573617775, | |
| "learning_rate": 1.01201956160768e-08, | |
| "loss": 0.1487, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.12491903454065323, | |
| "step": 4350, | |
| "valid_targets_mean": 2639.5, | |
| "valid_targets_min": 1146 | |
| }, | |
| { | |
| "epoch": 6.945773524720893, | |
| "grad_norm": 0.8297368312305293, | |
| "learning_rate": 7.74842793772601e-09, | |
| "loss": 0.1758, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.3191789388656616, | |
| "step": 4355, | |
| "valid_targets_mean": 4091.2, | |
| "valid_targets_min": 894 | |
| }, | |
| { | |
| "epoch": 6.953748006379586, | |
| "grad_norm": 0.5672862404897413, | |
| "learning_rate": 5.692820096054874e-09, | |
| "loss": 0.1712, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16838830709457397, | |
| "step": 4360, | |
| "valid_targets_mean": 3874.2, | |
| "valid_targets_min": 1136 | |
| }, | |
| { | |
| "epoch": 6.961722488038277, | |
| "grad_norm": 0.6673410547733448, | |
| "learning_rate": 3.9534045986888706e-09, | |
| "loss": 0.1445, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1458224207162857, | |
| "step": 4365, | |
| "valid_targets_mean": 2595.8, | |
| "valid_targets_min": 904 | |
| }, | |
| { | |
| "epoch": 6.96969696969697, | |
| "grad_norm": 0.6454669130283491, | |
| "learning_rate": 2.530208952953306e-09, | |
| "loss": 0.135, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.1391116827726364, | |
| "step": 4370, | |
| "valid_targets_mean": 2461.1, | |
| "valid_targets_min": 1391 | |
| }, | |
| { | |
| "epoch": 6.977671451355662, | |
| "grad_norm": 0.6808531645909409, | |
| "learning_rate": 1.4232556654314445e-09, | |
| "loss": 0.1572, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.16542094945907593, | |
| "step": 4375, | |
| "valid_targets_mean": 3267.7, | |
| "valid_targets_min": 1393 | |
| }, | |
| { | |
| "epoch": 6.985645933014354, | |
| "grad_norm": 0.6848879103739742, | |
| "learning_rate": 6.325622416136767e-10, | |
| "loss": 0.1422, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.15867075324058533, | |
| "step": 4380, | |
| "valid_targets_mean": 2476.5, | |
| "valid_targets_min": 649 | |
| }, | |
| { | |
| "epoch": 6.993620414673046, | |
| "grad_norm": 0.6367183670875135, | |
| "learning_rate": 1.581411856199644e-10, | |
| "loss": 0.1396, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.10416804254055023, | |
| "step": 4385, | |
| "valid_targets_mean": 2126.0, | |
| "valid_targets_min": 1296 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "loss_nan_ranks": 0, | |
| "loss_rank_avg": 0.13487885892391205, | |
| "step": 4389, | |
| "total_flos": 841177935118336.0, | |
| "train_loss": 0.10868970680736577, | |
| "train_runtime": 11861.0153, | |
| "train_samples_per_second": 5.919, | |
| "train_steps_per_second": 0.37, | |
| "valid_targets_mean": 2596.9, | |
| "valid_targets_min": 1254 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 4389, | |
| "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": 841177935118336.0, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |