Image-Text-to-Text
Transformers
Safetensors
qwen2_5_vl
image-editing
evaluation
judge
llm-as-judge
imgedit
conversational
text-generation-inference
Instructions to use bigshanedogg/ImgEdit-Judge with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use bigshanedogg/ImgEdit-Judge with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("image-text-to-text", model="bigshanedogg/ImgEdit-Judge") messages = [ { "role": "user", "content": [ {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"}, {"type": "text", "text": "What animal is on the candy?"} ] }, ] pipe(text=messages)# pip install -U transformers accelerate # Load model directly from transformers import AutoProcessor, AutoModelForMultimodalLM processor = AutoProcessor.from_pretrained("bigshanedogg/ImgEdit-Judge") model = AutoModelForMultimodalLM.from_pretrained("bigshanedogg/ImgEdit-Judge", device_map="auto") messages = [ { "role": "user", "content": [ {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"}, {"type": "text", "text": "What animal is on the candy?"} ] }, ] inputs = processor.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=256) print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:])) - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use bigshanedogg/ImgEdit-Judge with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "bigshanedogg/ImgEdit-Judge" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "bigshanedogg/ImgEdit-Judge", "messages": [ { "role": "user", "content": [ { "type": "text", "text": "Describe this image in one sentence." }, { "type": "image_url", "image_url": { "url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg" } } ] } ] }'Use Docker
docker model run hf.co/bigshanedogg/ImgEdit-Judge
- SGLang
How to use bigshanedogg/ImgEdit-Judge 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 "bigshanedogg/ImgEdit-Judge" \ --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": "bigshanedogg/ImgEdit-Judge", "messages": [ { "role": "user", "content": [ { "type": "text", "text": "Describe this image in one sentence." }, { "type": "image_url", "image_url": { "url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg" } } ] } ] }'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 "bigshanedogg/ImgEdit-Judge" \ --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": "bigshanedogg/ImgEdit-Judge", "messages": [ { "role": "user", "content": [ { "type": "text", "text": "Describe this image in one sentence." }, { "type": "image_url", "image_url": { "url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg" } } ] } ] }' - Docker Model Runner
How to use bigshanedogg/ImgEdit-Judge with Docker Model Runner:
docker model run hf.co/bigshanedogg/ImgEdit-Judge
Download trainer_state.json from bigshanedogg/ImgEdit-Judge: direct link, hf CLI and curl.
- Browser
- Download file 127 kB
-
https://huggingface.co/bigshanedogg/ImgEdit-Judge/resolve/main/trainer_state.json
- Command line
-
hf download hf://bigshanedogg/ImgEdit-Judge/trainer_state.json
-
curl -L -o trainer_state.json https://huggingface.co/bigshanedogg/ImgEdit-Judge/resolve/main/trainer_state.json
127 kB
| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 7243, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0013806433798149939, | |
| "grad_norm": 9589.8798828125, | |
| "learning_rate": 8.256880733944954e-09, | |
| "loss": 3.9054, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0027612867596299878, | |
| "grad_norm": 18668.462890625, | |
| "learning_rate": 1.743119266055046e-08, | |
| "loss": 3.9075, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.004141930139444981, | |
| "grad_norm": 16264.634765625, | |
| "learning_rate": 2.6605504587155964e-08, | |
| "loss": 3.5616, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0055225735192599755, | |
| "grad_norm": 9761.34765625, | |
| "learning_rate": 3.577981651376147e-08, | |
| "loss": 3.4741, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.006903216899074969, | |
| "grad_norm": 14939.7099609375, | |
| "learning_rate": 4.495412844036697e-08, | |
| "loss": 3.4493, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.008283860278889962, | |
| "grad_norm": 7365.0322265625, | |
| "learning_rate": 5.412844036697247e-08, | |
| "loss": 3.1691, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.009664503658704957, | |
| "grad_norm": 11767.228515625, | |
| "learning_rate": 6.330275229357797e-08, | |
| "loss": 3.0038, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.011045147038519951, | |
| "grad_norm": 1321.205810546875, | |
| "learning_rate": 7.247706422018348e-08, | |
| "loss": 2.3566, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.012425790418334944, | |
| "grad_norm": 3751.58544921875, | |
| "learning_rate": 8.165137614678899e-08, | |
| "loss": 2.2069, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.013806433798149938, | |
| "grad_norm": 624.3280029296875, | |
| "learning_rate": 9.08256880733945e-08, | |
| "loss": 2.2077, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.015187077177964931, | |
| "grad_norm": 4493.29150390625, | |
| "learning_rate": 1e-07, | |
| "loss": 2.1155, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.016567720557779924, | |
| "grad_norm": 3336.18798828125, | |
| "learning_rate": 1.0917431192660548e-07, | |
| "loss": 1.9459, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.01794836393759492, | |
| "grad_norm": 1115.7196044921875, | |
| "learning_rate": 1.18348623853211e-07, | |
| "loss": 1.8113, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.019329007317409913, | |
| "grad_norm": 92.8671875, | |
| "learning_rate": 1.275229357798165e-07, | |
| "loss": 1.7993, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.020709650697224906, | |
| "grad_norm": 778.4276123046875, | |
| "learning_rate": 1.36697247706422e-07, | |
| "loss": 1.7885, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.022090294077039902, | |
| "grad_norm": 798.2879028320312, | |
| "learning_rate": 1.4587155963302752e-07, | |
| "loss": 1.6825, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.023470937456854895, | |
| "grad_norm": 1144.1165771484375, | |
| "learning_rate": 1.5504587155963303e-07, | |
| "loss": 1.6023, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.024851580836669888, | |
| "grad_norm": 491.77020263671875, | |
| "learning_rate": 1.6422018348623852e-07, | |
| "loss": 1.6488, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.02623222421648488, | |
| "grad_norm": 689.5311889648438, | |
| "learning_rate": 1.7339449541284404e-07, | |
| "loss": 1.5284, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.027612867596299877, | |
| "grad_norm": 174.78677368164062, | |
| "learning_rate": 1.8256880733944953e-07, | |
| "loss": 1.5297, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.02899351097611487, | |
| "grad_norm": 86.76273345947266, | |
| "learning_rate": 1.9174311926605504e-07, | |
| "loss": 1.4364, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.030374154355929862, | |
| "grad_norm": 307.6069030761719, | |
| "learning_rate": 1.999999900005276e-07, | |
| "loss": 1.4365, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.031754797735744855, | |
| "grad_norm": 85.87425231933594, | |
| "learning_rate": 1.9999879006626124e-07, | |
| "loss": 1.3715, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.03313544111555985, | |
| "grad_norm": 5073.40625, | |
| "learning_rate": 1.9999559026501504e-07, | |
| "loss": 1.355, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.03451608449537485, | |
| "grad_norm": 92.59568786621094, | |
| "learning_rate": 1.9999039066078156e-07, | |
| "loss": 1.3226, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.03589672787518984, | |
| "grad_norm": 475.92718505859375, | |
| "learning_rate": 1.9998319135754725e-07, | |
| "loss": 1.2814, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.03727737125500483, | |
| "grad_norm": 187.81549072265625, | |
| "learning_rate": 1.9997399249929025e-07, | |
| "loss": 1.28, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.038658014634819826, | |
| "grad_norm": 75.67227935791016, | |
| "learning_rate": 1.999627942699778e-07, | |
| "loss": 1.2534, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.04003865801463482, | |
| "grad_norm": 531.5693969726562, | |
| "learning_rate": 1.9994959689356227e-07, | |
| "loss": 1.2658, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.04141930139444981, | |
| "grad_norm": 45.56630325317383, | |
| "learning_rate": 1.9993440063397683e-07, | |
| "loss": 1.2352, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.042799944774264805, | |
| "grad_norm": 38.359893798828125, | |
| "learning_rate": 1.9991720579513016e-07, | |
| "loss": 1.1872, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.044180588154079804, | |
| "grad_norm": 355.4065246582031, | |
| "learning_rate": 1.998980127209003e-07, | |
| "loss": 1.2015, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.0455612315338948, | |
| "grad_norm": 39.87456512451172, | |
| "learning_rate": 1.998768217951278e-07, | |
| "loss": 1.1813, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.04694187491370979, | |
| "grad_norm": 31.196317672729492, | |
| "learning_rate": 1.9985363344160817e-07, | |
| "loss": 1.1725, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.04832251829352478, | |
| "grad_norm": 1210.8070068359375, | |
| "learning_rate": 1.9982844812408322e-07, | |
| "loss": 1.1891, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.049703161673339775, | |
| "grad_norm": 79.10450744628906, | |
| "learning_rate": 1.9980126634623199e-07, | |
| "loss": 1.1979, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.05108380505315477, | |
| "grad_norm": 543.8863525390625, | |
| "learning_rate": 1.9977208865166032e-07, | |
| "loss": 1.1497, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.05246444843296976, | |
| "grad_norm": 399.2371520996094, | |
| "learning_rate": 1.9974091562389043e-07, | |
| "loss": 1.1484, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.05384509181278476, | |
| "grad_norm": 526.9529418945312, | |
| "learning_rate": 1.9970774788634892e-07, | |
| "loss": 1.1299, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.055225735192599754, | |
| "grad_norm": 25.764741897583008, | |
| "learning_rate": 1.9967258610235445e-07, | |
| "loss": 1.1383, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.056606378572414746, | |
| "grad_norm": 30.04519271850586, | |
| "learning_rate": 1.9963543097510443e-07, | |
| "loss": 1.1205, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.05798702195222974, | |
| "grad_norm": 24.468156814575195, | |
| "learning_rate": 1.9959628324766103e-07, | |
| "loss": 1.0978, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.05936766533204473, | |
| "grad_norm": 447.1950378417969, | |
| "learning_rate": 1.9955514370293613e-07, | |
| "loss": 1.1055, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.060748308711859725, | |
| "grad_norm": 20.887283325195312, | |
| "learning_rate": 1.9951201316367587e-07, | |
| "loss": 1.0812, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.06212895209167472, | |
| "grad_norm": 42.32823181152344, | |
| "learning_rate": 1.994668924924441e-07, | |
| "loss": 1.1052, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.06350959547148971, | |
| "grad_norm": 35.43871307373047, | |
| "learning_rate": 1.9941978259160516e-07, | |
| "loss": 1.0901, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.06489023885130471, | |
| "grad_norm": 541.15673828125, | |
| "learning_rate": 1.993706844033058e-07, | |
| "loss": 1.0796, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.0662708822311197, | |
| "grad_norm": 28.24175453186035, | |
| "learning_rate": 1.9931959890945632e-07, | |
| "loss": 1.0658, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.0676515256109347, | |
| "grad_norm": 30.67938804626465, | |
| "learning_rate": 1.9926652713171107e-07, | |
| "loss": 1.0795, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.0690321689907497, | |
| "grad_norm": 47.53831481933594, | |
| "learning_rate": 1.992114701314478e-07, | |
| "loss": 1.0375, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.07041281237056468, | |
| "grad_norm": 30.16608428955078, | |
| "learning_rate": 1.9915442900974658e-07, | |
| "loss": 1.0878, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.07179345575037968, | |
| "grad_norm": 57.50802230834961, | |
| "learning_rate": 1.9909540490736783e-07, | |
| "loss": 1.0704, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.07317409913019467, | |
| "grad_norm": 28.510169982910156, | |
| "learning_rate": 1.990343990047293e-07, | |
| "loss": 1.0603, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.07455474251000967, | |
| "grad_norm": 57.62672805786133, | |
| "learning_rate": 1.9897141252188275e-07, | |
| "loss": 1.0607, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.07593538588982465, | |
| "grad_norm": 36.105655670166016, | |
| "learning_rate": 1.9890644671848918e-07, | |
| "loss": 1.0538, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.07731602926963965, | |
| "grad_norm": 28.26181411743164, | |
| "learning_rate": 1.9883950289379406e-07, | |
| "loss": 1.0579, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.07869667264945465, | |
| "grad_norm": 24.388505935668945, | |
| "learning_rate": 1.9877058238660097e-07, | |
| "loss": 1.0203, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.08007731602926964, | |
| "grad_norm": 34.76478576660156, | |
| "learning_rate": 1.9869968657524506e-07, | |
| "loss": 1.0921, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.08145795940908464, | |
| "grad_norm": 45.64170455932617, | |
| "learning_rate": 1.986268168775654e-07, | |
| "loss": 1.048, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.08283860278889962, | |
| "grad_norm": 26.296384811401367, | |
| "learning_rate": 1.9855197475087674e-07, | |
| "loss": 1.0283, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.08421924616871462, | |
| "grad_norm": 36.26915740966797, | |
| "learning_rate": 1.9847516169194e-07, | |
| "loss": 1.0499, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.08559988954852961, | |
| "grad_norm": 26.26706886291504, | |
| "learning_rate": 1.9839637923693293e-07, | |
| "loss": 1.0349, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.08698053292834461, | |
| "grad_norm": 33.27248764038086, | |
| "learning_rate": 1.9831562896141878e-07, | |
| "loss": 1.0382, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.08836117630815961, | |
| "grad_norm": 23.02549171447754, | |
| "learning_rate": 1.9823291248031527e-07, | |
| "loss": 1.0236, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.0897418196879746, | |
| "grad_norm": 28.967567443847656, | |
| "learning_rate": 1.9814823144786195e-07, | |
| "loss": 1.0216, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.0911224630677896, | |
| "grad_norm": 24.4438419342041, | |
| "learning_rate": 1.9806158755758734e-07, | |
| "loss": 1.0368, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.09250310644760458, | |
| "grad_norm": 25.216018676757812, | |
| "learning_rate": 1.9797298254227495e-07, | |
| "loss": 1.0189, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.09388374982741958, | |
| "grad_norm": 16.15507698059082, | |
| "learning_rate": 1.9788241817392867e-07, | |
| "loss": 1.0594, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.09526439320723457, | |
| "grad_norm": 22.75001335144043, | |
| "learning_rate": 1.9778989626373733e-07, | |
| "loss": 0.9933, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.09664503658704957, | |
| "grad_norm": 25.44301414489746, | |
| "learning_rate": 1.9769541866203842e-07, | |
| "loss": 1.0272, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.09802567996686457, | |
| "grad_norm": 21.80268669128418, | |
| "learning_rate": 1.9759898725828115e-07, | |
| "loss": 1.013, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.09940632334667955, | |
| "grad_norm": 36.81167221069336, | |
| "learning_rate": 1.975006039809887e-07, | |
| "loss": 1.0306, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.10078696672649455, | |
| "grad_norm": 15.230224609375, | |
| "learning_rate": 1.974002707977195e-07, | |
| "loss": 1.0206, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.10216761010630954, | |
| "grad_norm": 49.497806549072266, | |
| "learning_rate": 1.9729798971502806e-07, | |
| "loss": 1.0412, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.10354825348612454, | |
| "grad_norm": 16.036041259765625, | |
| "learning_rate": 1.9719376277842476e-07, | |
| "loss": 1.0034, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.10492889686593952, | |
| "grad_norm": 26.95705795288086, | |
| "learning_rate": 1.9708759207233484e-07, | |
| "loss": 1.0365, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.10630954024575452, | |
| "grad_norm": 66.23104858398438, | |
| "learning_rate": 1.9697947972005694e-07, | |
| "loss": 1.0083, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.10769018362556952, | |
| "grad_norm": 24.981361389160156, | |
| "learning_rate": 1.9686942788372046e-07, | |
| "loss": 1.0252, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.10907082700538451, | |
| "grad_norm": 21.695451736450195, | |
| "learning_rate": 1.967574387642423e-07, | |
| "loss": 1.0149, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.11045147038519951, | |
| "grad_norm": 19.055261611938477, | |
| "learning_rate": 1.9664351460128302e-07, | |
| "loss": 0.9964, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.1118321137650145, | |
| "grad_norm": 86.02328491210938, | |
| "learning_rate": 1.965276576732019e-07, | |
| "loss": 1.0175, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.11321275714482949, | |
| "grad_norm": 25.227237701416016, | |
| "learning_rate": 1.9640987029701144e-07, | |
| "loss": 0.9969, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.11459340052464448, | |
| "grad_norm": 22.107818603515625, | |
| "learning_rate": 1.9629015482833098e-07, | |
| "loss": 1.0707, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.11597404390445948, | |
| "grad_norm": 18.990890502929688, | |
| "learning_rate": 1.961685136613396e-07, | |
| "loss": 1.0011, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.11735468728427448, | |
| "grad_norm": 26.50217628479004, | |
| "learning_rate": 1.9604494922872828e-07, | |
| "loss": 1.0267, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.11873533066408946, | |
| "grad_norm": 24.35108757019043, | |
| "learning_rate": 1.9591946400165117e-07, | |
| "loss": 1.0212, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.12011597404390446, | |
| "grad_norm": 17.32196044921875, | |
| "learning_rate": 1.957920604896763e-07, | |
| "loss": 1.0121, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.12149661742371945, | |
| "grad_norm": 21.199405670166016, | |
| "learning_rate": 1.9566274124073527e-07, | |
| "loss": 0.9838, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.12287726080353445, | |
| "grad_norm": 17.333036422729492, | |
| "learning_rate": 1.9553150884107227e-07, | |
| "loss": 0.9681, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.12425790418334944, | |
| "grad_norm": 14.428479194641113, | |
| "learning_rate": 1.9539836591519254e-07, | |
| "loss": 1.0182, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.12563854756316442, | |
| "grad_norm": 25.08713150024414, | |
| "learning_rate": 1.9526331512580972e-07, | |
| "loss": 1.0026, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.12701919094297942, | |
| "grad_norm": 31.190961837768555, | |
| "learning_rate": 1.9512635917379255e-07, | |
| "loss": 1.0008, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.12839983432279442, | |
| "grad_norm": 21.608415603637695, | |
| "learning_rate": 1.9498750079811111e-07, | |
| "loss": 1.0309, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.12978047770260942, | |
| "grad_norm": 24.187654495239258, | |
| "learning_rate": 1.948467427757818e-07, | |
| "loss": 1.002, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.13116112108242442, | |
| "grad_norm": 17.151939392089844, | |
| "learning_rate": 1.9470408792181181e-07, | |
| "loss": 0.9986, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.1325417644622394, | |
| "grad_norm": 20.126953125, | |
| "learning_rate": 1.9455953908914306e-07, | |
| "loss": 1.0028, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.1339224078420544, | |
| "grad_norm": 19.876928329467773, | |
| "learning_rate": 1.9441309916859486e-07, | |
| "loss": 1.0012, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.1353030512218694, | |
| "grad_norm": 16.082653045654297, | |
| "learning_rate": 1.942647710888063e-07, | |
| "loss": 1.0046, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.1366836946016844, | |
| "grad_norm": 16.94142723083496, | |
| "learning_rate": 1.9411455781617754e-07, | |
| "loss": 1.0032, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.1380643379814994, | |
| "grad_norm": 20.80314826965332, | |
| "learning_rate": 1.9396246235481054e-07, | |
| "loss": 0.9602, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.13944498136131436, | |
| "grad_norm": 27.991371154785156, | |
| "learning_rate": 1.9380848774644906e-07, | |
| "loss": 0.9983, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.14082562474112936, | |
| "grad_norm": 25.172107696533203, | |
| "learning_rate": 1.936526370704177e-07, | |
| "loss": 0.9849, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.14220626812094436, | |
| "grad_norm": 11.236927032470703, | |
| "learning_rate": 1.934949134435603e-07, | |
| "loss": 0.9842, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.14358691150075936, | |
| "grad_norm": 54.843788146972656, | |
| "learning_rate": 1.9333532002017789e-07, | |
| "loss": 1.0015, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.14496755488057433, | |
| "grad_norm": 23.669326782226562, | |
| "learning_rate": 1.9317385999196513e-07, | |
| "loss": 0.982, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.14634819826038933, | |
| "grad_norm": 17.040523529052734, | |
| "learning_rate": 1.93010536587947e-07, | |
| "loss": 0.9921, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.14772884164020433, | |
| "grad_norm": 17.456270217895508, | |
| "learning_rate": 1.9284535307441376e-07, | |
| "loss": 0.9958, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.14910948502001933, | |
| "grad_norm": 23.094526290893555, | |
| "learning_rate": 1.9267831275485598e-07, | |
| "loss": 0.9995, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.15049012839983433, | |
| "grad_norm": 17.812768936157227, | |
| "learning_rate": 1.9250941896989823e-07, | |
| "loss": 0.9906, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.1518707717796493, | |
| "grad_norm": 26.454042434692383, | |
| "learning_rate": 1.9233867509723242e-07, | |
| "loss": 0.9672, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.1532514151594643, | |
| "grad_norm": 33.25768280029297, | |
| "learning_rate": 1.9216608455155023e-07, | |
| "loss": 0.994, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.1546320585392793, | |
| "grad_norm": 41.439701080322266, | |
| "learning_rate": 1.919916507844747e-07, | |
| "loss": 0.9778, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.1560127019190943, | |
| "grad_norm": 34.070186614990234, | |
| "learning_rate": 1.9181537728449144e-07, | |
| "loss": 0.9868, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.1573933452989093, | |
| "grad_norm": 13.763279914855957, | |
| "learning_rate": 1.9163726757687854e-07, | |
| "loss": 0.9916, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.15877398867872428, | |
| "grad_norm": 19.878803253173828, | |
| "learning_rate": 1.9145732522363634e-07, | |
| "loss": 0.9852, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.16015463205853928, | |
| "grad_norm": 16.75155258178711, | |
| "learning_rate": 1.912755538234161e-07, | |
| "loss": 0.9802, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.16153527543835428, | |
| "grad_norm": 15.308502197265625, | |
| "learning_rate": 1.9109195701144803e-07, | |
| "loss": 0.9624, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.16291591881816928, | |
| "grad_norm": 15.146383285522461, | |
| "learning_rate": 1.9090653845946854e-07, | |
| "loss": 0.9593, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.16429656219798425, | |
| "grad_norm": 18.412540435791016, | |
| "learning_rate": 1.9071930187564692e-07, | |
| "loss": 0.9898, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.16567720557779925, | |
| "grad_norm": 15.691420555114746, | |
| "learning_rate": 1.9053025100451104e-07, | |
| "loss": 0.9436, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.16705784895761425, | |
| "grad_norm": 15.704176902770996, | |
| "learning_rate": 1.903393896268727e-07, | |
| "loss": 0.9716, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.16843849233742925, | |
| "grad_norm": 16.53687858581543, | |
| "learning_rate": 1.9014672155975165e-07, | |
| "loss": 0.9605, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.16981913571724425, | |
| "grad_norm": 14.696842193603516, | |
| "learning_rate": 1.8995225065629962e-07, | |
| "loss": 0.9724, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.17119977909705922, | |
| "grad_norm": 13.728157043457031, | |
| "learning_rate": 1.8975598080572302e-07, | |
| "loss": 0.9907, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.17258042247687422, | |
| "grad_norm": 17.215879440307617, | |
| "learning_rate": 1.8955791593320528e-07, | |
| "loss": 0.9635, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.17396106585668922, | |
| "grad_norm": 23.80732536315918, | |
| "learning_rate": 1.893580599998283e-07, | |
| "loss": 0.9676, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.17534170923650422, | |
| "grad_norm": 24.319040298461914, | |
| "learning_rate": 1.891564170024933e-07, | |
| "loss": 0.9898, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.17672235261631922, | |
| "grad_norm": 25.764873504638672, | |
| "learning_rate": 1.889529909738408e-07, | |
| "loss": 0.954, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.1781029959961342, | |
| "grad_norm": 15.074317932128906, | |
| "learning_rate": 1.887477859821699e-07, | |
| "loss": 0.9644, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.1794836393759492, | |
| "grad_norm": 19.326274871826172, | |
| "learning_rate": 1.8854080613135724e-07, | |
| "loss": 0.966, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.1808642827557642, | |
| "grad_norm": 17.36568832397461, | |
| "learning_rate": 1.8833205556077455e-07, | |
| "loss": 0.9704, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.1822449261355792, | |
| "grad_norm": 16.688398361206055, | |
| "learning_rate": 1.8812153844520604e-07, | |
| "loss": 0.9532, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.18362556951539416, | |
| "grad_norm": 22.168596267700195, | |
| "learning_rate": 1.8790925899476497e-07, | |
| "loss": 0.9653, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.18500621289520916, | |
| "grad_norm": 19.795974731445312, | |
| "learning_rate": 1.8769522145480936e-07, | |
| "loss": 0.9567, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.18638685627502416, | |
| "grad_norm": 18.69921112060547, | |
| "learning_rate": 1.8747943010585702e-07, | |
| "loss": 0.9774, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.18776749965483916, | |
| "grad_norm": 18.926239013671875, | |
| "learning_rate": 1.8726188926350014e-07, | |
| "loss": 0.9507, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.18914814303465416, | |
| "grad_norm": 20.12693977355957, | |
| "learning_rate": 1.8704260327831886e-07, | |
| "loss": 0.9829, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.19052878641446913, | |
| "grad_norm": 12.830967903137207, | |
| "learning_rate": 1.868215765357942e-07, | |
| "loss": 0.9586, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.19190942979428413, | |
| "grad_norm": 16.826265335083008, | |
| "learning_rate": 1.8659881345622052e-07, | |
| "loss": 0.9434, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.19329007317409913, | |
| "grad_norm": 15.19537353515625, | |
| "learning_rate": 1.86374318494617e-07, | |
| "loss": 0.9872, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.19467071655391413, | |
| "grad_norm": 16.144060134887695, | |
| "learning_rate": 1.8614809614063848e-07, | |
| "loss": 0.9585, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.19605135993372913, | |
| "grad_norm": 25.445268630981445, | |
| "learning_rate": 1.8592015091848597e-07, | |
| "loss": 0.9232, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.1974320033135441, | |
| "grad_norm": 18.136150360107422, | |
| "learning_rate": 1.8569048738681584e-07, | |
| "loss": 0.9399, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.1988126466933591, | |
| "grad_norm": 15.041813850402832, | |
| "learning_rate": 1.8545911013864873e-07, | |
| "loss": 0.9348, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.2001932900731741, | |
| "grad_norm": 11.633391380310059, | |
| "learning_rate": 1.8522602380127784e-07, | |
| "loss": 0.9647, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.2015739334529891, | |
| "grad_norm": 18.411531448364258, | |
| "learning_rate": 1.849912330361763e-07, | |
| "loss": 0.9439, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.20295457683280407, | |
| "grad_norm": 21.11094856262207, | |
| "learning_rate": 1.8475474253890391e-07, | |
| "loss": 0.9766, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.20433522021261907, | |
| "grad_norm": 32.33837127685547, | |
| "learning_rate": 1.8451655703901322e-07, | |
| "loss": 0.9665, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.20571586359243407, | |
| "grad_norm": 24.007678985595703, | |
| "learning_rate": 1.84276681299955e-07, | |
| "loss": 0.9604, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.20709650697224907, | |
| "grad_norm": 19.237699508666992, | |
| "learning_rate": 1.8403512011898307e-07, | |
| "loss": 0.9723, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.20847715035206407, | |
| "grad_norm": 19.735424041748047, | |
| "learning_rate": 1.8379187832705812e-07, | |
| "loss": 0.9532, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.20985779373187904, | |
| "grad_norm": 15.690709114074707, | |
| "learning_rate": 1.8354696078875135e-07, | |
| "loss": 0.9535, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.21123843711169404, | |
| "grad_norm": 18.543996810913086, | |
| "learning_rate": 1.8330037240214692e-07, | |
| "loss": 0.9393, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 0.21261908049150904, | |
| "grad_norm": 17.197307586669922, | |
| "learning_rate": 1.830521180987443e-07, | |
| "loss": 0.9523, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.21399972387132404, | |
| "grad_norm": 11.924728393554688, | |
| "learning_rate": 1.828022028433593e-07, | |
| "loss": 0.9377, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.21538036725113904, | |
| "grad_norm": 16.14398956298828, | |
| "learning_rate": 1.8255063163402511e-07, | |
| "loss": 0.9635, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.21676101063095402, | |
| "grad_norm": 17.4932804107666, | |
| "learning_rate": 1.8229740950189218e-07, | |
| "loss": 0.9628, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 0.21814165401076901, | |
| "grad_norm": 15.191761016845703, | |
| "learning_rate": 1.820425415111276e-07, | |
| "loss": 0.923, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.21952229739058401, | |
| "grad_norm": 22.885875701904297, | |
| "learning_rate": 1.8178603275881375e-07, | |
| "loss": 0.938, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 0.22090294077039901, | |
| "grad_norm": 12.818890571594238, | |
| "learning_rate": 1.815278883748466e-07, | |
| "loss": 0.9666, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.222283584150214, | |
| "grad_norm": 13.444008827209473, | |
| "learning_rate": 1.8126811352183288e-07, | |
| "loss": 0.9431, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 0.223664227530029, | |
| "grad_norm": 24.44390106201172, | |
| "learning_rate": 1.81006713394987e-07, | |
| "loss": 0.9715, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.22504487090984399, | |
| "grad_norm": 11.453673362731934, | |
| "learning_rate": 1.80743693222027e-07, | |
| "loss": 0.9552, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 0.22642551428965899, | |
| "grad_norm": 19.372905731201172, | |
| "learning_rate": 1.8047905826307007e-07, | |
| "loss": 0.9316, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.22780615766947399, | |
| "grad_norm": 14.868914604187012, | |
| "learning_rate": 1.8021281381052748e-07, | |
| "loss": 0.9565, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.22918680104928896, | |
| "grad_norm": 14.215982437133789, | |
| "learning_rate": 1.7994496518899853e-07, | |
| "loss": 0.9421, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.23056744442910396, | |
| "grad_norm": 14.750982284545898, | |
| "learning_rate": 1.7967551775516415e-07, | |
| "loss": 0.926, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 0.23194808780891896, | |
| "grad_norm": 12.635498046875, | |
| "learning_rate": 1.7940447689767982e-07, | |
| "loss": 0.939, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.23332873118873396, | |
| "grad_norm": 21.397735595703125, | |
| "learning_rate": 1.791318480370677e-07, | |
| "loss": 0.9399, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 0.23470937456854896, | |
| "grad_norm": 17.604341506958008, | |
| "learning_rate": 1.7885763662560832e-07, | |
| "loss": 0.9601, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.23609001794836393, | |
| "grad_norm": 32.83750915527344, | |
| "learning_rate": 1.7858184814723157e-07, | |
| "loss": 0.9526, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 0.23747066132817893, | |
| "grad_norm": 1420.344482421875, | |
| "learning_rate": 1.7830448811740682e-07, | |
| "loss": 0.9593, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.23885130470799393, | |
| "grad_norm": 16.57451057434082, | |
| "learning_rate": 1.7802556208303287e-07, | |
| "loss": 0.9221, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 0.24023194808780893, | |
| "grad_norm": 20.10437774658203, | |
| "learning_rate": 1.7774507562232687e-07, | |
| "loss": 0.9514, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.2416125914676239, | |
| "grad_norm": 22.54353904724121, | |
| "learning_rate": 1.774630343447128e-07, | |
| "loss": 0.9502, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.2429932348474389, | |
| "grad_norm": 17.310949325561523, | |
| "learning_rate": 1.7717944389070932e-07, | |
| "loss": 0.9634, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.2443738782272539, | |
| "grad_norm": 21.16713523864746, | |
| "learning_rate": 1.7689430993181682e-07, | |
| "loss": 0.9317, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 0.2457545216070689, | |
| "grad_norm": 12.144413948059082, | |
| "learning_rate": 1.7660763817040426e-07, | |
| "loss": 0.9393, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.2471351649868839, | |
| "grad_norm": 11.153665542602539, | |
| "learning_rate": 1.7631943433959488e-07, | |
| "loss": 0.9287, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 0.24851580836669887, | |
| "grad_norm": 19.016427993774414, | |
| "learning_rate": 1.7602970420315165e-07, | |
| "loss": 0.9454, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.24989645174651387, | |
| "grad_norm": 20.053916931152344, | |
| "learning_rate": 1.75738453555362e-07, | |
| "loss": 0.9459, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 0.25127709512632884, | |
| "grad_norm": 14.30462646484375, | |
| "learning_rate": 1.7544568822092197e-07, | |
| "loss": 0.9144, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.25265773850614387, | |
| "grad_norm": 11.231522560119629, | |
| "learning_rate": 1.7515141405481963e-07, | |
| "loss": 0.924, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 0.25403838188595884, | |
| "grad_norm": 17.285886764526367, | |
| "learning_rate": 1.748556369422181e-07, | |
| "loss": 0.9503, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.25541902526577387, | |
| "grad_norm": 17.03244972229004, | |
| "learning_rate": 1.7455836279833772e-07, | |
| "loss": 0.9603, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.25679966864558884, | |
| "grad_norm": 13.772926330566406, | |
| "learning_rate": 1.742595975683379e-07, | |
| "loss": 0.9276, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.2581803120254038, | |
| "grad_norm": 19.941713333129883, | |
| "learning_rate": 1.7395934722719808e-07, | |
| "loss": 0.9377, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 0.25956095540521884, | |
| "grad_norm": 14.966099739074707, | |
| "learning_rate": 1.7365761777959834e-07, | |
| "loss": 0.9526, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 0.2609415987850338, | |
| "grad_norm": 16.90435791015625, | |
| "learning_rate": 1.733544152597993e-07, | |
| "loss": 0.9513, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 0.26232224216484884, | |
| "grad_norm": 21.27311897277832, | |
| "learning_rate": 1.7304974573152143e-07, | |
| "loss": 0.9089, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.2637028855446638, | |
| "grad_norm": 10.289287567138672, | |
| "learning_rate": 1.727436152878237e-07, | |
| "loss": 0.9366, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 0.2650835289244788, | |
| "grad_norm": 13.82613754272461, | |
| "learning_rate": 1.7243603005098185e-07, | |
| "loss": 0.9248, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.2664641723042938, | |
| "grad_norm": 14.732545852661133, | |
| "learning_rate": 1.721269961723659e-07, | |
| "loss": 0.9361, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 0.2678448156841088, | |
| "grad_norm": 11.531627655029297, | |
| "learning_rate": 1.7181651983231716e-07, | |
| "loss": 0.9297, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 0.2692254590639238, | |
| "grad_norm": 17.648893356323242, | |
| "learning_rate": 1.7150460724002452e-07, | |
| "loss": 0.9659, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.2706061024437388, | |
| "grad_norm": 17.644508361816406, | |
| "learning_rate": 1.7119126463340038e-07, | |
| "loss": 0.9118, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 0.27198674582355375, | |
| "grad_norm": 21.6595458984375, | |
| "learning_rate": 1.7087649827895594e-07, | |
| "loss": 0.9324, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 0.2733673892033688, | |
| "grad_norm": 10.022526741027832, | |
| "learning_rate": 1.7056031447167566e-07, | |
| "loss": 0.9666, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 0.27474803258318375, | |
| "grad_norm": 19.683359146118164, | |
| "learning_rate": 1.702427195348917e-07, | |
| "loss": 0.9154, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 0.2761286759629988, | |
| "grad_norm": 11.023941993713379, | |
| "learning_rate": 1.6992371982015717e-07, | |
| "loss": 0.9382, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.27750931934281375, | |
| "grad_norm": 13.107163429260254, | |
| "learning_rate": 1.6960332170711918e-07, | |
| "loss": 0.9532, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 0.2788899627226287, | |
| "grad_norm": 17.789451599121094, | |
| "learning_rate": 1.6928153160339132e-07, | |
| "loss": 0.9285, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 0.28027060610244375, | |
| "grad_norm": 17.33866310119629, | |
| "learning_rate": 1.6895835594442552e-07, | |
| "loss": 0.9278, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 0.2816512494822587, | |
| "grad_norm": 12.742042541503906, | |
| "learning_rate": 1.6863380119338323e-07, | |
| "loss": 0.9518, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 0.28303189286207375, | |
| "grad_norm": 13.640534400939941, | |
| "learning_rate": 1.6830787384100632e-07, | |
| "loss": 0.9365, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.2844125362418887, | |
| "grad_norm": 16.454971313476562, | |
| "learning_rate": 1.6798058040548715e-07, | |
| "loss": 0.9566, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 0.2857931796217037, | |
| "grad_norm": 19.262805938720703, | |
| "learning_rate": 1.6765192743233821e-07, | |
| "loss": 0.9181, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 0.2871738230015187, | |
| "grad_norm": 19.85011863708496, | |
| "learning_rate": 1.6732192149426138e-07, | |
| "loss": 0.9353, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 0.2885544663813337, | |
| "grad_norm": 11.263802528381348, | |
| "learning_rate": 1.669905691910163e-07, | |
| "loss": 0.9188, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 0.28993510976114867, | |
| "grad_norm": 15.778621673583984, | |
| "learning_rate": 1.6665787714928835e-07, | |
| "loss": 0.9354, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.2913157531409637, | |
| "grad_norm": 12.294133186340332, | |
| "learning_rate": 1.6632385202255645e-07, | |
| "loss": 0.9314, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 0.29269639652077867, | |
| "grad_norm": 19.767330169677734, | |
| "learning_rate": 1.6598850049095954e-07, | |
| "loss": 0.9344, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 0.2940770399005937, | |
| "grad_norm": 13.18829345703125, | |
| "learning_rate": 1.656518292611634e-07, | |
| "loss": 0.9277, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 0.29545768328040867, | |
| "grad_norm": 15.715974807739258, | |
| "learning_rate": 1.6531384506622613e-07, | |
| "loss": 0.9796, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 0.29683832666022364, | |
| "grad_norm": 16.956928253173828, | |
| "learning_rate": 1.6497455466546397e-07, | |
| "loss": 0.9015, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.29821897004003867, | |
| "grad_norm": 11.247440338134766, | |
| "learning_rate": 1.6463396484431563e-07, | |
| "loss": 0.9337, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 0.29959961341985364, | |
| "grad_norm": 21.892221450805664, | |
| "learning_rate": 1.6429208241420685e-07, | |
| "loss": 0.9317, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 0.30098025679966867, | |
| "grad_norm": 17.855491638183594, | |
| "learning_rate": 1.6394891421241428e-07, | |
| "loss": 0.9155, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 0.30236090017948364, | |
| "grad_norm": 13.436010360717773, | |
| "learning_rate": 1.6360446710192846e-07, | |
| "loss": 0.9472, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 0.3037415435592986, | |
| "grad_norm": 9.978911399841309, | |
| "learning_rate": 1.6325874797131674e-07, | |
| "loss": 0.9487, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.30512218693911364, | |
| "grad_norm": 20.90780258178711, | |
| "learning_rate": 1.6291176373458558e-07, | |
| "loss": 0.949, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 0.3065028303189286, | |
| "grad_norm": 14.61563491821289, | |
| "learning_rate": 1.6256352133104206e-07, | |
| "loss": 0.9018, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 0.30788347369874364, | |
| "grad_norm": 13.13970947265625, | |
| "learning_rate": 1.622140277251553e-07, | |
| "loss": 0.9327, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 0.3092641170785586, | |
| "grad_norm": 19.569936752319336, | |
| "learning_rate": 1.618632899064171e-07, | |
| "loss": 0.945, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 0.3106447604583736, | |
| "grad_norm": 17.013219833374023, | |
| "learning_rate": 1.6151131488920214e-07, | |
| "loss": 0.912, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.3120254038381886, | |
| "grad_norm": 15.479214668273926, | |
| "learning_rate": 1.6115810971262774e-07, | |
| "loss": 0.9268, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 0.3134060472180036, | |
| "grad_norm": 9.596595764160156, | |
| "learning_rate": 1.6080368144041303e-07, | |
| "loss": 0.9237, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 0.3147866905978186, | |
| "grad_norm": 17.678857803344727, | |
| "learning_rate": 1.6044803716073783e-07, | |
| "loss": 0.9508, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 0.3161673339776336, | |
| "grad_norm": 13.671900749206543, | |
| "learning_rate": 1.6009118398610065e-07, | |
| "loss": 0.9366, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 0.31754797735744855, | |
| "grad_norm": 20.82811737060547, | |
| "learning_rate": 1.5973312905317666e-07, | |
| "loss": 0.9223, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.3189286207372636, | |
| "grad_norm": 12.565200805664062, | |
| "learning_rate": 1.5937387952267488e-07, | |
| "loss": 0.9377, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 0.32030926411707855, | |
| "grad_norm": 20.529130935668945, | |
| "learning_rate": 1.5901344257919497e-07, | |
| "loss": 0.9164, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 0.3216899074968936, | |
| "grad_norm": 11.798916816711426, | |
| "learning_rate": 1.5865182543108359e-07, | |
| "loss": 0.8963, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 0.32307055087670855, | |
| "grad_norm": 16.206594467163086, | |
| "learning_rate": 1.582890353102901e-07, | |
| "loss": 0.9477, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 0.3244511942565235, | |
| "grad_norm": 13.417816162109375, | |
| "learning_rate": 1.579250794722222e-07, | |
| "loss": 0.9125, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.32583183763633855, | |
| "grad_norm": 10.749960899353027, | |
| "learning_rate": 1.5755996519560055e-07, | |
| "loss": 0.9306, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 0.3272124810161535, | |
| "grad_norm": 20.59378433227539, | |
| "learning_rate": 1.5719369978231337e-07, | |
| "loss": 0.9186, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 0.3285931243959685, | |
| "grad_norm": 15.732232093811035, | |
| "learning_rate": 1.5682629055727033e-07, | |
| "loss": 0.9225, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 0.3299737677757835, | |
| "grad_norm": 23.157899856567383, | |
| "learning_rate": 1.564577448682561e-07, | |
| "loss": 0.8965, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 0.3313544111555985, | |
| "grad_norm": 12.062108039855957, | |
| "learning_rate": 1.5608807008578345e-07, | |
| "loss": 0.9193, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.3327350545354135, | |
| "grad_norm": 15.707069396972656, | |
| "learning_rate": 1.5571727360294564e-07, | |
| "loss": 0.9003, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 0.3341156979152285, | |
| "grad_norm": 15.863463401794434, | |
| "learning_rate": 1.553453628352689e-07, | |
| "loss": 0.9204, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 0.33549634129504347, | |
| "grad_norm": 12.108427047729492, | |
| "learning_rate": 1.5497234522056389e-07, | |
| "loss": 0.9188, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 0.3368769846748585, | |
| "grad_norm": 17.726287841796875, | |
| "learning_rate": 1.5459822821877683e-07, | |
| "loss": 0.9133, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 0.33825762805467346, | |
| "grad_norm": 13.311707496643066, | |
| "learning_rate": 1.542230193118408e-07, | |
| "loss": 0.9053, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.3396382714344885, | |
| "grad_norm": 14.009614944458008, | |
| "learning_rate": 1.5384672600352548e-07, | |
| "loss": 0.901, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 0.34101891481430346, | |
| "grad_norm": 17.43226432800293, | |
| "learning_rate": 1.5346935581928764e-07, | |
| "loss": 0.8888, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 0.34239955819411844, | |
| "grad_norm": 18.421289443969727, | |
| "learning_rate": 1.5309091630612027e-07, | |
| "loss": 0.9291, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 0.34378020157393346, | |
| "grad_norm": 19.394193649291992, | |
| "learning_rate": 1.527114150324018e-07, | |
| "loss": 0.9068, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 0.34516084495374844, | |
| "grad_norm": 12.585472106933594, | |
| "learning_rate": 1.5233085958774473e-07, | |
| "loss": 0.9359, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.34654148833356346, | |
| "grad_norm": 14.959007263183594, | |
| "learning_rate": 1.5194925758284383e-07, | |
| "loss": 0.9156, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 0.34792213171337844, | |
| "grad_norm": 12.919416427612305, | |
| "learning_rate": 1.5156661664932387e-07, | |
| "loss": 0.9252, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 0.3493027750931934, | |
| "grad_norm": 21.312864303588867, | |
| "learning_rate": 1.5118294443958722e-07, | |
| "loss": 0.9073, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 0.35068341847300843, | |
| "grad_norm": 17.94431495666504, | |
| "learning_rate": 1.5079824862666044e-07, | |
| "loss": 0.9167, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 0.3520640618528234, | |
| "grad_norm": 11.176458358764648, | |
| "learning_rate": 1.5041253690404125e-07, | |
| "loss": 0.9044, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.35344470523263843, | |
| "grad_norm": 12.279413223266602, | |
| "learning_rate": 1.5002581698554432e-07, | |
| "loss": 0.9171, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 0.3548253486124534, | |
| "grad_norm": 11.652252197265625, | |
| "learning_rate": 1.496380966051472e-07, | |
| "loss": 0.9201, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 0.3562059919922684, | |
| "grad_norm": 23.9465274810791, | |
| "learning_rate": 1.4924938351683547e-07, | |
| "loss": 0.9073, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 0.3575866353720834, | |
| "grad_norm": 20.31412696838379, | |
| "learning_rate": 1.4885968549444804e-07, | |
| "loss": 0.9127, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 0.3589672787518984, | |
| "grad_norm": 12.084100723266602, | |
| "learning_rate": 1.4846901033152117e-07, | |
| "loss": 0.9176, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.3603479221317134, | |
| "grad_norm": 32.15177536010742, | |
| "learning_rate": 1.4807736584113303e-07, | |
| "loss": 0.894, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 0.3617285655115284, | |
| "grad_norm": 10.340327262878418, | |
| "learning_rate": 1.4768475985574718e-07, | |
| "loss": 0.8991, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 0.36310920889134335, | |
| "grad_norm": 16.143047332763672, | |
| "learning_rate": 1.4729120022705616e-07, | |
| "loss": 0.903, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 0.3644898522711584, | |
| "grad_norm": 15.663701057434082, | |
| "learning_rate": 1.4689669482582417e-07, | |
| "loss": 0.8975, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 0.36587049565097335, | |
| "grad_norm": 13.738362312316895, | |
| "learning_rate": 1.4650125154172996e-07, | |
| "loss": 0.9014, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.3672511390307883, | |
| "grad_norm": 17.67761993408203, | |
| "learning_rate": 1.4610487828320888e-07, | |
| "loss": 0.8811, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 0.36863178241060335, | |
| "grad_norm": 13.131311416625977, | |
| "learning_rate": 1.457075829772948e-07, | |
| "loss": 0.9531, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 0.3700124257904183, | |
| "grad_norm": 12.904136657714844, | |
| "learning_rate": 1.453093735694614e-07, | |
| "loss": 0.909, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 0.37139306917023335, | |
| "grad_norm": 10.439740180969238, | |
| "learning_rate": 1.4491025802346358e-07, | |
| "loss": 0.9112, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 0.3727737125500483, | |
| "grad_norm": 11.147518157958984, | |
| "learning_rate": 1.445102443211779e-07, | |
| "loss": 0.9141, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.3741543559298633, | |
| "grad_norm": 13.87960147857666, | |
| "learning_rate": 1.441093404624431e-07, | |
| "loss": 0.9037, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 0.3755349993096783, | |
| "grad_norm": 11.555074691772461, | |
| "learning_rate": 1.4370755446490007e-07, | |
| "loss": 0.8855, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 0.3769156426894933, | |
| "grad_norm": 11.576875686645508, | |
| "learning_rate": 1.433048943638315e-07, | |
| "loss": 0.9418, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 0.3782962860693083, | |
| "grad_norm": 19.2021427154541, | |
| "learning_rate": 1.4290136821200128e-07, | |
| "loss": 0.9048, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 0.3796769294491233, | |
| "grad_norm": 13.335321426391602, | |
| "learning_rate": 1.424969840794933e-07, | |
| "loss": 0.9005, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.38105757282893826, | |
| "grad_norm": 12.150760650634766, | |
| "learning_rate": 1.4209175005355013e-07, | |
| "loss": 0.9189, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 0.3824382162087533, | |
| "grad_norm": 14.57982063293457, | |
| "learning_rate": 1.4168567423841134e-07, | |
| "loss": 0.9088, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 0.38381885958856826, | |
| "grad_norm": 19.238052368164062, | |
| "learning_rate": 1.4127876475515128e-07, | |
| "loss": 0.907, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 0.3851995029683833, | |
| "grad_norm": 14.475991249084473, | |
| "learning_rate": 1.4087102974151683e-07, | |
| "loss": 0.8976, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 0.38658014634819826, | |
| "grad_norm": 17.70484161376953, | |
| "learning_rate": 1.4046247735176456e-07, | |
| "loss": 0.9104, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.38796078972801323, | |
| "grad_norm": 12.97494125366211, | |
| "learning_rate": 1.4005311575649766e-07, | |
| "loss": 0.8907, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 0.38934143310782826, | |
| "grad_norm": 10.209972381591797, | |
| "learning_rate": 1.3964295314250256e-07, | |
| "loss": 0.8926, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 0.39072207648764323, | |
| "grad_norm": 11.831448554992676, | |
| "learning_rate": 1.3923199771258518e-07, | |
| "loss": 0.9268, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 0.39210271986745826, | |
| "grad_norm": 13.091449737548828, | |
| "learning_rate": 1.388202576854069e-07, | |
| "loss": 0.9105, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 0.39348336324727323, | |
| "grad_norm": 14.359809875488281, | |
| "learning_rate": 1.3840774129532023e-07, | |
| "loss": 0.8722, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.3948640066270882, | |
| "grad_norm": 14.059985160827637, | |
| "learning_rate": 1.3799445679220403e-07, | |
| "loss": 0.9125, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 0.39624465000690323, | |
| "grad_norm": 13.30297565460205, | |
| "learning_rate": 1.375804124412986e-07, | |
| "loss": 0.9037, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 0.3976252933867182, | |
| "grad_norm": 18.505708694458008, | |
| "learning_rate": 1.3716561652304042e-07, | |
| "loss": 0.909, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 0.39900593676653323, | |
| "grad_norm": 10.83780288696289, | |
| "learning_rate": 1.3675007733289646e-07, | |
| "loss": 0.8903, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 0.4003865801463482, | |
| "grad_norm": 15.617920875549316, | |
| "learning_rate": 1.363338031811983e-07, | |
| "loss": 0.9052, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.4017672235261632, | |
| "grad_norm": 11.155744552612305, | |
| "learning_rate": 1.35916802392976e-07, | |
| "loss": 0.9349, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 0.4031478669059782, | |
| "grad_norm": 14.732954978942871, | |
| "learning_rate": 1.3549908330779147e-07, | |
| "loss": 0.9025, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 0.4045285102857932, | |
| "grad_norm": 13.338746070861816, | |
| "learning_rate": 1.3508065427957189e-07, | |
| "loss": 0.9227, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 0.40590915366560815, | |
| "grad_norm": 13.729426383972168, | |
| "learning_rate": 1.3466152367644253e-07, | |
| "loss": 0.924, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 0.4072897970454232, | |
| "grad_norm": 16.463890075683594, | |
| "learning_rate": 1.3424169988055924e-07, | |
| "loss": 0.9331, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.40867044042523815, | |
| "grad_norm": 15.352555274963379, | |
| "learning_rate": 1.3382119128794112e-07, | |
| "loss": 0.9168, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 0.4100510838050532, | |
| "grad_norm": 17.169523239135742, | |
| "learning_rate": 1.334000063083024e-07, | |
| "loss": 0.9143, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 0.41143172718486815, | |
| "grad_norm": 14.708706855773926, | |
| "learning_rate": 1.329781533648844e-07, | |
| "loss": 0.9239, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 0.4128123705646831, | |
| "grad_norm": 18.491050720214844, | |
| "learning_rate": 1.3255564089428677e-07, | |
| "loss": 0.8911, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 0.41419301394449815, | |
| "grad_norm": 14.142806053161621, | |
| "learning_rate": 1.3213247734629927e-07, | |
| "loss": 0.9103, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.4155736573243131, | |
| "grad_norm": 12.782434463500977, | |
| "learning_rate": 1.3170867118373233e-07, | |
| "loss": 0.8891, | |
| "step": 3010 | |
| }, | |
| { | |
| "epoch": 0.41695430070412814, | |
| "grad_norm": 16.018287658691406, | |
| "learning_rate": 1.3128423088224793e-07, | |
| "loss": 0.8862, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 0.4183349440839431, | |
| "grad_norm": 21.54560089111328, | |
| "learning_rate": 1.308591649301903e-07, | |
| "loss": 0.8928, | |
| "step": 3030 | |
| }, | |
| { | |
| "epoch": 0.4197155874637581, | |
| "grad_norm": 11.521778106689453, | |
| "learning_rate": 1.3043348182841584e-07, | |
| "loss": 0.9047, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 0.4210962308435731, | |
| "grad_norm": 19.725364685058594, | |
| "learning_rate": 1.3000719009012344e-07, | |
| "loss": 0.8959, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.4224768742233881, | |
| "grad_norm": 13.046162605285645, | |
| "learning_rate": 1.2958029824068385e-07, | |
| "loss": 0.9058, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 0.4238575176032031, | |
| "grad_norm": 13.915671348571777, | |
| "learning_rate": 1.291528148174696e-07, | |
| "loss": 0.894, | |
| "step": 3070 | |
| }, | |
| { | |
| "epoch": 0.4252381609830181, | |
| "grad_norm": 12.897782325744629, | |
| "learning_rate": 1.2872474836968387e-07, | |
| "loss": 0.9059, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 0.42661880436283306, | |
| "grad_norm": 10.964375495910645, | |
| "learning_rate": 1.282961074581899e-07, | |
| "loss": 0.8967, | |
| "step": 3090 | |
| }, | |
| { | |
| "epoch": 0.4279994477426481, | |
| "grad_norm": 9.513810157775879, | |
| "learning_rate": 1.278669006553393e-07, | |
| "loss": 0.9054, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.42938009112246306, | |
| "grad_norm": 12.593969345092773, | |
| "learning_rate": 1.2743713654480118e-07, | |
| "loss": 0.9081, | |
| "step": 3110 | |
| }, | |
| { | |
| "epoch": 0.4307607345022781, | |
| "grad_norm": 14.793880462646484, | |
| "learning_rate": 1.2700682372139002e-07, | |
| "loss": 0.8769, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 0.43214137788209306, | |
| "grad_norm": 16.92717933654785, | |
| "learning_rate": 1.2657597079089402e-07, | |
| "loss": 0.886, | |
| "step": 3130 | |
| }, | |
| { | |
| "epoch": 0.43352202126190803, | |
| "grad_norm": 18.84389305114746, | |
| "learning_rate": 1.261445863699029e-07, | |
| "loss": 0.8938, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 0.43490266464172306, | |
| "grad_norm": 9.070972442626953, | |
| "learning_rate": 1.2571267908563568e-07, | |
| "loss": 0.8916, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.43628330802153803, | |
| "grad_norm": 23.482677459716797, | |
| "learning_rate": 1.25280257575768e-07, | |
| "loss": 0.8734, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 0.43766395140135306, | |
| "grad_norm": 11.570613861083984, | |
| "learning_rate": 1.2484733048825948e-07, | |
| "loss": 0.9176, | |
| "step": 3170 | |
| }, | |
| { | |
| "epoch": 0.43904459478116803, | |
| "grad_norm": 10.780313491821289, | |
| "learning_rate": 1.2441390648118078e-07, | |
| "loss": 0.9035, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 0.440425238160983, | |
| "grad_norm": 15.462409973144531, | |
| "learning_rate": 1.2397999422254036e-07, | |
| "loss": 0.8994, | |
| "step": 3190 | |
| }, | |
| { | |
| "epoch": 0.44180588154079803, | |
| "grad_norm": 16.559764862060547, | |
| "learning_rate": 1.2354560239011116e-07, | |
| "loss": 0.8983, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.443186524920613, | |
| "grad_norm": 12.226709365844727, | |
| "learning_rate": 1.2311073967125715e-07, | |
| "loss": 0.9093, | |
| "step": 3210 | |
| }, | |
| { | |
| "epoch": 0.444567168300428, | |
| "grad_norm": 15.039121627807617, | |
| "learning_rate": 1.226754147627595e-07, | |
| "loss": 0.8938, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 0.445947811680243, | |
| "grad_norm": 14.54739761352539, | |
| "learning_rate": 1.222396363706426e-07, | |
| "loss": 0.9145, | |
| "step": 3230 | |
| }, | |
| { | |
| "epoch": 0.447328455060058, | |
| "grad_norm": 10.211729049682617, | |
| "learning_rate": 1.218034132100001e-07, | |
| "loss": 0.8998, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 0.448709098439873, | |
| "grad_norm": 15.60191822052002, | |
| "learning_rate": 1.2136675400482045e-07, | |
| "loss": 0.8949, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.45008974181968797, | |
| "grad_norm": 19.02963638305664, | |
| "learning_rate": 1.2092966748781265e-07, | |
| "loss": 0.8796, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 0.45147038519950294, | |
| "grad_norm": 12.375081062316895, | |
| "learning_rate": 1.2049216240023132e-07, | |
| "loss": 0.8977, | |
| "step": 3270 | |
| }, | |
| { | |
| "epoch": 0.45285102857931797, | |
| "grad_norm": 10.026701927185059, | |
| "learning_rate": 1.2005424749170214e-07, | |
| "loss": 0.8927, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 0.45423167195913294, | |
| "grad_norm": 13.246793746948242, | |
| "learning_rate": 1.1961593152004672e-07, | |
| "loss": 0.9048, | |
| "step": 3290 | |
| }, | |
| { | |
| "epoch": 0.45561231533894797, | |
| "grad_norm": 16.54412269592285, | |
| "learning_rate": 1.1917722325110751e-07, | |
| "loss": 0.9152, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.45699295871876294, | |
| "grad_norm": 16.543588638305664, | |
| "learning_rate": 1.1873813145857248e-07, | |
| "loss": 0.881, | |
| "step": 3310 | |
| }, | |
| { | |
| "epoch": 0.4583736020985779, | |
| "grad_norm": 14.020906448364258, | |
| "learning_rate": 1.182986649237996e-07, | |
| "loss": 0.8833, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 0.45975424547839294, | |
| "grad_norm": 10.389933586120605, | |
| "learning_rate": 1.178588324356414e-07, | |
| "loss": 0.9167, | |
| "step": 3330 | |
| }, | |
| { | |
| "epoch": 0.4611348888582079, | |
| "grad_norm": 34.92580032348633, | |
| "learning_rate": 1.1741864279026898e-07, | |
| "loss": 0.8895, | |
| "step": 3340 | |
| }, | |
| { | |
| "epoch": 0.46251553223802294, | |
| "grad_norm": 14.951874732971191, | |
| "learning_rate": 1.169781047909962e-07, | |
| "loss": 0.8962, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.4638961756178379, | |
| "grad_norm": 13.5904541015625, | |
| "learning_rate": 1.1653722724810364e-07, | |
| "loss": 0.9038, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 0.4652768189976529, | |
| "grad_norm": 12.143257141113281, | |
| "learning_rate": 1.1609601897866239e-07, | |
| "loss": 0.8975, | |
| "step": 3370 | |
| }, | |
| { | |
| "epoch": 0.4666574623774679, | |
| "grad_norm": 11.456350326538086, | |
| "learning_rate": 1.1565448880635773e-07, | |
| "loss": 0.8859, | |
| "step": 3380 | |
| }, | |
| { | |
| "epoch": 0.4680381057572829, | |
| "grad_norm": 15.442255020141602, | |
| "learning_rate": 1.1521264556131258e-07, | |
| "loss": 0.9243, | |
| "step": 3390 | |
| }, | |
| { | |
| "epoch": 0.4694187491370979, | |
| "grad_norm": 16.532827377319336, | |
| "learning_rate": 1.1477049807991104e-07, | |
| "loss": 0.9195, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.4707993925169129, | |
| "grad_norm": 16.205148696899414, | |
| "learning_rate": 1.1432805520462156e-07, | |
| "loss": 0.9012, | |
| "step": 3410 | |
| }, | |
| { | |
| "epoch": 0.47218003589672786, | |
| "grad_norm": 10.40243148803711, | |
| "learning_rate": 1.1388532578382019e-07, | |
| "loss": 0.8975, | |
| "step": 3420 | |
| }, | |
| { | |
| "epoch": 0.4735606792765429, | |
| "grad_norm": 10.075249671936035, | |
| "learning_rate": 1.1344231867161356e-07, | |
| "loss": 0.8838, | |
| "step": 3430 | |
| }, | |
| { | |
| "epoch": 0.47494132265635786, | |
| "grad_norm": 13.057167053222656, | |
| "learning_rate": 1.1299904272766178e-07, | |
| "loss": 0.9195, | |
| "step": 3440 | |
| }, | |
| { | |
| "epoch": 0.4763219660361729, | |
| "grad_norm": 8.801608085632324, | |
| "learning_rate": 1.125555068170014e-07, | |
| "loss": 0.8992, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.47770260941598786, | |
| "grad_norm": 17.592405319213867, | |
| "learning_rate": 1.1211171980986791e-07, | |
| "loss": 0.8886, | |
| "step": 3460 | |
| }, | |
| { | |
| "epoch": 0.4790832527958028, | |
| "grad_norm": 14.348983764648438, | |
| "learning_rate": 1.1166769058151855e-07, | |
| "loss": 0.8861, | |
| "step": 3470 | |
| }, | |
| { | |
| "epoch": 0.48046389617561786, | |
| "grad_norm": 13.283276557922363, | |
| "learning_rate": 1.1122342801205465e-07, | |
| "loss": 0.906, | |
| "step": 3480 | |
| }, | |
| { | |
| "epoch": 0.4818445395554328, | |
| "grad_norm": 14.10513687133789, | |
| "learning_rate": 1.1077894098624416e-07, | |
| "loss": 0.8895, | |
| "step": 3490 | |
| }, | |
| { | |
| "epoch": 0.4832251829352478, | |
| "grad_norm": 18.94774627685547, | |
| "learning_rate": 1.1033423839334392e-07, | |
| "loss": 0.9052, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.4846058263150628, | |
| "grad_norm": 15.77293586730957, | |
| "learning_rate": 1.0988932912692179e-07, | |
| "loss": 0.9007, | |
| "step": 3510 | |
| }, | |
| { | |
| "epoch": 0.4859864696948778, | |
| "grad_norm": 12.630675315856934, | |
| "learning_rate": 1.0944422208467898e-07, | |
| "loss": 0.875, | |
| "step": 3520 | |
| }, | |
| { | |
| "epoch": 0.4873671130746928, | |
| "grad_norm": 18.209169387817383, | |
| "learning_rate": 1.0899892616827195e-07, | |
| "loss": 0.9095, | |
| "step": 3530 | |
| }, | |
| { | |
| "epoch": 0.4887477564545078, | |
| "grad_norm": 13.609612464904785, | |
| "learning_rate": 1.0855345028313444e-07, | |
| "loss": 0.9058, | |
| "step": 3540 | |
| }, | |
| { | |
| "epoch": 0.49012839983432277, | |
| "grad_norm": 14.771902084350586, | |
| "learning_rate": 1.0810780333829937e-07, | |
| "loss": 0.8762, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.4915090432141378, | |
| "grad_norm": 16.29448890686035, | |
| "learning_rate": 1.0766199424622069e-07, | |
| "loss": 0.9025, | |
| "step": 3560 | |
| }, | |
| { | |
| "epoch": 0.49288968659395277, | |
| "grad_norm": 13.09625244140625, | |
| "learning_rate": 1.0721603192259505e-07, | |
| "loss": 0.883, | |
| "step": 3570 | |
| }, | |
| { | |
| "epoch": 0.4942703299737678, | |
| "grad_norm": 20.347095489501953, | |
| "learning_rate": 1.0676992528618368e-07, | |
| "loss": 0.8895, | |
| "step": 3580 | |
| }, | |
| { | |
| "epoch": 0.49565097335358277, | |
| "grad_norm": 14.950133323669434, | |
| "learning_rate": 1.0632368325863379e-07, | |
| "loss": 0.884, | |
| "step": 3590 | |
| }, | |
| { | |
| "epoch": 0.49703161673339774, | |
| "grad_norm": 13.336871147155762, | |
| "learning_rate": 1.0587731476430033e-07, | |
| "loss": 0.8923, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.49841226011321277, | |
| "grad_norm": 8.866421699523926, | |
| "learning_rate": 1.0543082873006746e-07, | |
| "loss": 0.8869, | |
| "step": 3610 | |
| }, | |
| { | |
| "epoch": 0.49979290349302774, | |
| "grad_norm": 10.176736831665039, | |
| "learning_rate": 1.0498423408516994e-07, | |
| "loss": 0.8838, | |
| "step": 3620 | |
| }, | |
| { | |
| "epoch": 0.5011735468728428, | |
| "grad_norm": 8.377656936645508, | |
| "learning_rate": 1.0453753976101474e-07, | |
| "loss": 0.8945, | |
| "step": 3630 | |
| }, | |
| { | |
| "epoch": 0.5025541902526577, | |
| "grad_norm": 16.51427459716797, | |
| "learning_rate": 1.0409075469100217e-07, | |
| "loss": 0.8841, | |
| "step": 3640 | |
| }, | |
| { | |
| "epoch": 0.5039348336324727, | |
| "grad_norm": 10.73915958404541, | |
| "learning_rate": 1.0364388781034747e-07, | |
| "loss": 0.916, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.5053154770122877, | |
| "grad_norm": 8.6974515914917, | |
| "learning_rate": 1.0319694805590192e-07, | |
| "loss": 0.8848, | |
| "step": 3660 | |
| }, | |
| { | |
| "epoch": 0.5066961203921028, | |
| "grad_norm": 10.280749320983887, | |
| "learning_rate": 1.0274994436597428e-07, | |
| "loss": 0.8749, | |
| "step": 3670 | |
| }, | |
| { | |
| "epoch": 0.5080767637719177, | |
| "grad_norm": 12.423198699951172, | |
| "learning_rate": 1.0230288568015189e-07, | |
| "loss": 0.9051, | |
| "step": 3680 | |
| }, | |
| { | |
| "epoch": 0.5094574071517327, | |
| "grad_norm": 14.206807136535645, | |
| "learning_rate": 1.0185578093912195e-07, | |
| "loss": 0.8956, | |
| "step": 3690 | |
| }, | |
| { | |
| "epoch": 0.5108380505315477, | |
| "grad_norm": 14.779512405395508, | |
| "learning_rate": 1.0140863908449275e-07, | |
| "loss": 0.8976, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.5122186939113627, | |
| "grad_norm": 13.520800590515137, | |
| "learning_rate": 1.009614690586148e-07, | |
| "loss": 0.9017, | |
| "step": 3710 | |
| }, | |
| { | |
| "epoch": 0.5135993372911777, | |
| "grad_norm": 11.862271308898926, | |
| "learning_rate": 1.0051427980440195e-07, | |
| "loss": 0.8938, | |
| "step": 3720 | |
| }, | |
| { | |
| "epoch": 0.5149799806709927, | |
| "grad_norm": 9.344632148742676, | |
| "learning_rate": 1.0006708026515264e-07, | |
| "loss": 0.8974, | |
| "step": 3730 | |
| }, | |
| { | |
| "epoch": 0.5163606240508076, | |
| "grad_norm": 13.059805870056152, | |
| "learning_rate": 9.961987938437107e-08, | |
| "loss": 0.9056, | |
| "step": 3740 | |
| }, | |
| { | |
| "epoch": 0.5177412674306227, | |
| "grad_norm": 13.468391418457031, | |
| "learning_rate": 9.917268610558816e-08, | |
| "loss": 0.8825, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.5191219108104377, | |
| "grad_norm": 19.47498321533203, | |
| "learning_rate": 9.872550937218281e-08, | |
| "loss": 0.8951, | |
| "step": 3760 | |
| }, | |
| { | |
| "epoch": 0.5205025541902527, | |
| "grad_norm": 17.413328170776367, | |
| "learning_rate": 9.827835812720307e-08, | |
| "loss": 0.8724, | |
| "step": 3770 | |
| }, | |
| { | |
| "epoch": 0.5218831975700676, | |
| "grad_norm": 13.719565391540527, | |
| "learning_rate": 9.783124131318724e-08, | |
| "loss": 0.909, | |
| "step": 3780 | |
| }, | |
| { | |
| "epoch": 0.5232638409498827, | |
| "grad_norm": 10.409643173217773, | |
| "learning_rate": 9.738416787198507e-08, | |
| "loss": 0.8806, | |
| "step": 3790 | |
| }, | |
| { | |
| "epoch": 0.5246444843296977, | |
| "grad_norm": 11.702856063842773, | |
| "learning_rate": 9.693714674457884e-08, | |
| "loss": 0.8901, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.5260251277095126, | |
| "grad_norm": 11.778882026672363, | |
| "learning_rate": 9.649018687090461e-08, | |
| "loss": 0.8999, | |
| "step": 3810 | |
| }, | |
| { | |
| "epoch": 0.5274057710893276, | |
| "grad_norm": 9.282418251037598, | |
| "learning_rate": 9.60432971896735e-08, | |
| "loss": 0.9042, | |
| "step": 3820 | |
| }, | |
| { | |
| "epoch": 0.5287864144691427, | |
| "grad_norm": 14.665790557861328, | |
| "learning_rate": 9.559648663819283e-08, | |
| "loss": 0.8909, | |
| "step": 3830 | |
| }, | |
| { | |
| "epoch": 0.5301670578489576, | |
| "grad_norm": 9.734098434448242, | |
| "learning_rate": 9.514976415218732e-08, | |
| "loss": 0.8778, | |
| "step": 3840 | |
| }, | |
| { | |
| "epoch": 0.5315477012287726, | |
| "grad_norm": 10.756970405578613, | |
| "learning_rate": 9.47031386656206e-08, | |
| "loss": 0.8627, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.5329283446085876, | |
| "grad_norm": 9.799631118774414, | |
| "learning_rate": 9.425661911051641e-08, | |
| "loss": 0.8874, | |
| "step": 3860 | |
| }, | |
| { | |
| "epoch": 0.5343089879884027, | |
| "grad_norm": 9.511898040771484, | |
| "learning_rate": 9.38102144167799e-08, | |
| "loss": 0.8786, | |
| "step": 3870 | |
| }, | |
| { | |
| "epoch": 0.5356896313682176, | |
| "grad_norm": 10.795829772949219, | |
| "learning_rate": 9.336393351201914e-08, | |
| "loss": 0.9222, | |
| "step": 3880 | |
| }, | |
| { | |
| "epoch": 0.5370702747480326, | |
| "grad_norm": 9.76403522491455, | |
| "learning_rate": 9.29177853213666e-08, | |
| "loss": 0.8907, | |
| "step": 3890 | |
| }, | |
| { | |
| "epoch": 0.5384509181278476, | |
| "grad_norm": 17.731040954589844, | |
| "learning_rate": 9.247177876730053e-08, | |
| "loss": 0.882, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.5398315615076625, | |
| "grad_norm": 9.753355026245117, | |
| "learning_rate": 9.202592276946674e-08, | |
| "loss": 0.8926, | |
| "step": 3910 | |
| }, | |
| { | |
| "epoch": 0.5412122048874776, | |
| "grad_norm": 20.043432235717773, | |
| "learning_rate": 9.158022624449991e-08, | |
| "loss": 0.8926, | |
| "step": 3920 | |
| }, | |
| { | |
| "epoch": 0.5425928482672926, | |
| "grad_norm": 15.967229843139648, | |
| "learning_rate": 9.11346981058456e-08, | |
| "loss": 0.9006, | |
| "step": 3930 | |
| }, | |
| { | |
| "epoch": 0.5439734916471075, | |
| "grad_norm": 10.734868049621582, | |
| "learning_rate": 9.068934726358165e-08, | |
| "loss": 0.8909, | |
| "step": 3940 | |
| }, | |
| { | |
| "epoch": 0.5453541350269225, | |
| "grad_norm": 11.771434783935547, | |
| "learning_rate": 9.024418262424034e-08, | |
| "loss": 0.8781, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.5467347784067376, | |
| "grad_norm": 17.062509536743164, | |
| "learning_rate": 8.979921309062998e-08, | |
| "loss": 0.8726, | |
| "step": 3960 | |
| }, | |
| { | |
| "epoch": 0.5481154217865525, | |
| "grad_norm": 12.76464557647705, | |
| "learning_rate": 8.935444756165705e-08, | |
| "loss": 0.8977, | |
| "step": 3970 | |
| }, | |
| { | |
| "epoch": 0.5494960651663675, | |
| "grad_norm": 14.029791831970215, | |
| "learning_rate": 8.890989493214805e-08, | |
| "loss": 0.8768, | |
| "step": 3980 | |
| }, | |
| { | |
| "epoch": 0.5508767085461825, | |
| "grad_norm": 13.004902839660645, | |
| "learning_rate": 8.84655640926719e-08, | |
| "loss": 0.911, | |
| "step": 3990 | |
| }, | |
| { | |
| "epoch": 0.5522573519259976, | |
| "grad_norm": 13.776290893554688, | |
| "learning_rate": 8.802146392936176e-08, | |
| "loss": 0.872, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.5536379953058125, | |
| "grad_norm": 14.18698501586914, | |
| "learning_rate": 8.757760332373768e-08, | |
| "loss": 0.8712, | |
| "step": 4010 | |
| }, | |
| { | |
| "epoch": 0.5550186386856275, | |
| "grad_norm": 11.412870407104492, | |
| "learning_rate": 8.713399115252869e-08, | |
| "loss": 0.8836, | |
| "step": 4020 | |
| }, | |
| { | |
| "epoch": 0.5563992820654425, | |
| "grad_norm": 10.589845657348633, | |
| "learning_rate": 8.669063628749552e-08, | |
| "loss": 0.8835, | |
| "step": 4030 | |
| }, | |
| { | |
| "epoch": 0.5577799254452575, | |
| "grad_norm": 10.09622859954834, | |
| "learning_rate": 8.624754759525296e-08, | |
| "loss": 0.8882, | |
| "step": 4040 | |
| }, | |
| { | |
| "epoch": 0.5591605688250725, | |
| "grad_norm": 11.894771575927734, | |
| "learning_rate": 8.580473393709274e-08, | |
| "loss": 0.8831, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 0.5605412122048875, | |
| "grad_norm": 15.063697814941406, | |
| "learning_rate": 8.536220416880602e-08, | |
| "loss": 0.8883, | |
| "step": 4060 | |
| }, | |
| { | |
| "epoch": 0.5619218555847024, | |
| "grad_norm": 18.955522537231445, | |
| "learning_rate": 8.49199671405067e-08, | |
| "loss": 0.8998, | |
| "step": 4070 | |
| }, | |
| { | |
| "epoch": 0.5633024989645175, | |
| "grad_norm": 12.944504737854004, | |
| "learning_rate": 8.447803169645404e-08, | |
| "loss": 0.9048, | |
| "step": 4080 | |
| }, | |
| { | |
| "epoch": 0.5646831423443325, | |
| "grad_norm": 9.729756355285645, | |
| "learning_rate": 8.403640667487603e-08, | |
| "loss": 0.8779, | |
| "step": 4090 | |
| }, | |
| { | |
| "epoch": 0.5660637857241475, | |
| "grad_norm": 11.885720252990723, | |
| "learning_rate": 8.359510090779247e-08, | |
| "loss": 0.8924, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.5674444291039624, | |
| "grad_norm": 9.443900108337402, | |
| "learning_rate": 8.315412322083848e-08, | |
| "loss": 0.8771, | |
| "step": 4110 | |
| }, | |
| { | |
| "epoch": 0.5688250724837774, | |
| "grad_norm": 8.952618598937988, | |
| "learning_rate": 8.271348243308794e-08, | |
| "loss": 0.8868, | |
| "step": 4120 | |
| }, | |
| { | |
| "epoch": 0.5702057158635925, | |
| "grad_norm": 13.090829849243164, | |
| "learning_rate": 8.227318735687707e-08, | |
| "loss": 0.879, | |
| "step": 4130 | |
| }, | |
| { | |
| "epoch": 0.5715863592434074, | |
| "grad_norm": 11.320798873901367, | |
| "learning_rate": 8.183324679762821e-08, | |
| "loss": 0.8685, | |
| "step": 4140 | |
| }, | |
| { | |
| "epoch": 0.5729670026232224, | |
| "grad_norm": 11.434670448303223, | |
| "learning_rate": 8.139366955367384e-08, | |
| "loss": 0.9166, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 0.5743476460030374, | |
| "grad_norm": 11.928689956665039, | |
| "learning_rate": 8.095446441608048e-08, | |
| "loss": 0.9255, | |
| "step": 4160 | |
| }, | |
| { | |
| "epoch": 0.5757282893828524, | |
| "grad_norm": 12.919870376586914, | |
| "learning_rate": 8.051564016847292e-08, | |
| "loss": 0.8856, | |
| "step": 4170 | |
| }, | |
| { | |
| "epoch": 0.5771089327626674, | |
| "grad_norm": 9.934977531433105, | |
| "learning_rate": 8.007720558685853e-08, | |
| "loss": 0.8894, | |
| "step": 4180 | |
| }, | |
| { | |
| "epoch": 0.5784895761424824, | |
| "grad_norm": 11.866926193237305, | |
| "learning_rate": 7.963916943945184e-08, | |
| "loss": 0.9074, | |
| "step": 4190 | |
| }, | |
| { | |
| "epoch": 0.5798702195222973, | |
| "grad_norm": 18.457015991210938, | |
| "learning_rate": 7.920154048649918e-08, | |
| "loss": 0.8909, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.5812508629021124, | |
| "grad_norm": 14.267375946044922, | |
| "learning_rate": 7.876432748010333e-08, | |
| "loss": 0.891, | |
| "step": 4210 | |
| }, | |
| { | |
| "epoch": 0.5826315062819274, | |
| "grad_norm": 21.52684211730957, | |
| "learning_rate": 7.83275391640486e-08, | |
| "loss": 0.8798, | |
| "step": 4220 | |
| }, | |
| { | |
| "epoch": 0.5840121496617424, | |
| "grad_norm": 20.638736724853516, | |
| "learning_rate": 7.789118427362597e-08, | |
| "loss": 0.9008, | |
| "step": 4230 | |
| }, | |
| { | |
| "epoch": 0.5853927930415573, | |
| "grad_norm": 21.53960609436035, | |
| "learning_rate": 7.745527153545849e-08, | |
| "loss": 0.889, | |
| "step": 4240 | |
| }, | |
| { | |
| "epoch": 0.5867734364213724, | |
| "grad_norm": 14.670390129089355, | |
| "learning_rate": 7.701980966732646e-08, | |
| "loss": 0.9003, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 0.5881540798011874, | |
| "grad_norm": 9.364107131958008, | |
| "learning_rate": 7.658480737799344e-08, | |
| "loss": 0.8963, | |
| "step": 4260 | |
| }, | |
| { | |
| "epoch": 0.5895347231810023, | |
| "grad_norm": 15.018786430358887, | |
| "learning_rate": 7.615027336703171e-08, | |
| "loss": 0.8606, | |
| "step": 4270 | |
| }, | |
| { | |
| "epoch": 0.5909153665608173, | |
| "grad_norm": 147.060546875, | |
| "learning_rate": 7.571621632464877e-08, | |
| "loss": 0.8993, | |
| "step": 4280 | |
| }, | |
| { | |
| "epoch": 0.5922960099406324, | |
| "grad_norm": 7.983867645263672, | |
| "learning_rate": 7.5282644931513e-08, | |
| "loss": 0.8941, | |
| "step": 4290 | |
| }, | |
| { | |
| "epoch": 0.5936766533204473, | |
| "grad_norm": 16.283424377441406, | |
| "learning_rate": 7.484956785858052e-08, | |
| "loss": 0.8877, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.5950572967002623, | |
| "grad_norm": 11.41619873046875, | |
| "learning_rate": 7.44169937669214e-08, | |
| "loss": 0.8838, | |
| "step": 4310 | |
| }, | |
| { | |
| "epoch": 0.5964379400800773, | |
| "grad_norm": 16.697721481323242, | |
| "learning_rate": 7.398493130754684e-08, | |
| "loss": 0.9022, | |
| "step": 4320 | |
| }, | |
| { | |
| "epoch": 0.5978185834598924, | |
| "grad_norm": 12.903840065002441, | |
| "learning_rate": 7.355338912123579e-08, | |
| "loss": 0.8927, | |
| "step": 4330 | |
| }, | |
| { | |
| "epoch": 0.5991992268397073, | |
| "grad_norm": 13.159421920776367, | |
| "learning_rate": 7.312237583836238e-08, | |
| "loss": 0.8979, | |
| "step": 4340 | |
| }, | |
| { | |
| "epoch": 0.6005798702195223, | |
| "grad_norm": 10.291422843933105, | |
| "learning_rate": 7.269190007872318e-08, | |
| "loss": 0.8845, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 0.6019605135993373, | |
| "grad_norm": 10.45628547668457, | |
| "learning_rate": 7.226197045136502e-08, | |
| "loss": 0.8788, | |
| "step": 4360 | |
| }, | |
| { | |
| "epoch": 0.6033411569791522, | |
| "grad_norm": 18.41653823852539, | |
| "learning_rate": 7.183259555441248e-08, | |
| "loss": 0.9034, | |
| "step": 4370 | |
| }, | |
| { | |
| "epoch": 0.6047218003589673, | |
| "grad_norm": 16.795162200927734, | |
| "learning_rate": 7.140378397489633e-08, | |
| "loss": 0.8748, | |
| "step": 4380 | |
| }, | |
| { | |
| "epoch": 0.6061024437387823, | |
| "grad_norm": 12.827390670776367, | |
| "learning_rate": 7.097554428858144e-08, | |
| "loss": 0.8849, | |
| "step": 4390 | |
| }, | |
| { | |
| "epoch": 0.6074830871185972, | |
| "grad_norm": 10.136775970458984, | |
| "learning_rate": 7.054788505979556e-08, | |
| "loss": 0.8914, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.6088637304984122, | |
| "grad_norm": 16.414024353027344, | |
| "learning_rate": 7.012081484125787e-08, | |
| "loss": 0.8614, | |
| "step": 4410 | |
| }, | |
| { | |
| "epoch": 0.6102443738782273, | |
| "grad_norm": 8.86361026763916, | |
| "learning_rate": 6.9694342173908e-08, | |
| "loss": 0.9051, | |
| "step": 4420 | |
| }, | |
| { | |
| "epoch": 0.6116250172580423, | |
| "grad_norm": 10.634222030639648, | |
| "learning_rate": 6.926847558673514e-08, | |
| "loss": 0.9006, | |
| "step": 4430 | |
| }, | |
| { | |
| "epoch": 0.6130056606378572, | |
| "grad_norm": 11.486374855041504, | |
| "learning_rate": 6.884322359660769e-08, | |
| "loss": 0.8729, | |
| "step": 4440 | |
| }, | |
| { | |
| "epoch": 0.6143863040176722, | |
| "grad_norm": 12.955568313598633, | |
| "learning_rate": 6.84185947081026e-08, | |
| "loss": 0.8946, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 0.6157669473974873, | |
| "grad_norm": 11.915677070617676, | |
| "learning_rate": 6.79945974133356e-08, | |
| "loss": 0.8912, | |
| "step": 4460 | |
| }, | |
| { | |
| "epoch": 0.6171475907773022, | |
| "grad_norm": 1072.02490234375, | |
| "learning_rate": 6.757124019179113e-08, | |
| "loss": 0.8867, | |
| "step": 4470 | |
| }, | |
| { | |
| "epoch": 0.6185282341571172, | |
| "grad_norm": 14.126385688781738, | |
| "learning_rate": 6.714853151015289e-08, | |
| "loss": 0.8758, | |
| "step": 4480 | |
| }, | |
| { | |
| "epoch": 0.6199088775369322, | |
| "grad_norm": 12.1957368850708, | |
| "learning_rate": 6.672647982213464e-08, | |
| "loss": 0.9049, | |
| "step": 4490 | |
| }, | |
| { | |
| "epoch": 0.6212895209167472, | |
| "grad_norm": 8.881195068359375, | |
| "learning_rate": 6.630509356831074e-08, | |
| "loss": 0.884, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.6226701642965622, | |
| "grad_norm": 18.463071823120117, | |
| "learning_rate": 6.588438117594772e-08, | |
| "loss": 0.8859, | |
| "step": 4510 | |
| }, | |
| { | |
| "epoch": 0.6240508076763772, | |
| "grad_norm": 17.912851333618164, | |
| "learning_rate": 6.546435105883558e-08, | |
| "loss": 0.8523, | |
| "step": 4520 | |
| }, | |
| { | |
| "epoch": 0.6254314510561921, | |
| "grad_norm": 11.206585884094238, | |
| "learning_rate": 6.504501161711964e-08, | |
| "loss": 0.9026, | |
| "step": 4530 | |
| }, | |
| { | |
| "epoch": 0.6268120944360072, | |
| "grad_norm": 16.016130447387695, | |
| "learning_rate": 6.462637123713236e-08, | |
| "loss": 0.8887, | |
| "step": 4540 | |
| }, | |
| { | |
| "epoch": 0.6281927378158222, | |
| "grad_norm": 17.54102897644043, | |
| "learning_rate": 6.420843829122573e-08, | |
| "loss": 0.8769, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 0.6295733811956372, | |
| "grad_norm": 9.402512550354004, | |
| "learning_rate": 6.379122113760389e-08, | |
| "loss": 0.8883, | |
| "step": 4560 | |
| }, | |
| { | |
| "epoch": 0.6309540245754521, | |
| "grad_norm": 9.921586990356445, | |
| "learning_rate": 6.337472812015591e-08, | |
| "loss": 0.9053, | |
| "step": 4570 | |
| }, | |
| { | |
| "epoch": 0.6323346679552672, | |
| "grad_norm": 19.98529624938965, | |
| "learning_rate": 6.295896756828888e-08, | |
| "loss": 0.8686, | |
| "step": 4580 | |
| }, | |
| { | |
| "epoch": 0.6337153113350822, | |
| "grad_norm": 10.525534629821777, | |
| "learning_rate": 6.254394779676138e-08, | |
| "loss": 0.8954, | |
| "step": 4590 | |
| }, | |
| { | |
| "epoch": 0.6350959547148971, | |
| "grad_norm": 17.254629135131836, | |
| "learning_rate": 6.212967710551719e-08, | |
| "loss": 0.8745, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.6364765980947121, | |
| "grad_norm": 16.953174591064453, | |
| "learning_rate": 6.17161637795194e-08, | |
| "loss": 0.8894, | |
| "step": 4610 | |
| }, | |
| { | |
| "epoch": 0.6378572414745272, | |
| "grad_norm": 21.23188018798828, | |
| "learning_rate": 6.130341608858444e-08, | |
| "loss": 0.8889, | |
| "step": 4620 | |
| }, | |
| { | |
| "epoch": 0.6392378848543421, | |
| "grad_norm": 12.853032112121582, | |
| "learning_rate": 6.0891442287217e-08, | |
| "loss": 0.9051, | |
| "step": 4630 | |
| }, | |
| { | |
| "epoch": 0.6406185282341571, | |
| "grad_norm": 12.195560455322266, | |
| "learning_rate": 6.048025061444474e-08, | |
| "loss": 0.8803, | |
| "step": 4640 | |
| }, | |
| { | |
| "epoch": 0.6419991716139721, | |
| "grad_norm": 55.15872573852539, | |
| "learning_rate": 6.006984929365377e-08, | |
| "loss": 0.8842, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 0.6433798149937872, | |
| "grad_norm": 9.930737495422363, | |
| "learning_rate": 5.966024653242379e-08, | |
| "loss": 0.8956, | |
| "step": 4660 | |
| }, | |
| { | |
| "epoch": 0.6447604583736021, | |
| "grad_norm": 8.529269218444824, | |
| "learning_rate": 5.9251450522364276e-08, | |
| "loss": 0.8836, | |
| "step": 4670 | |
| }, | |
| { | |
| "epoch": 0.6461411017534171, | |
| "grad_norm": 10.782485008239746, | |
| "learning_rate": 5.88434694389506e-08, | |
| "loss": 0.8871, | |
| "step": 4680 | |
| }, | |
| { | |
| "epoch": 0.6475217451332321, | |
| "grad_norm": 8.634961128234863, | |
| "learning_rate": 5.843631144136048e-08, | |
| "loss": 0.8659, | |
| "step": 4690 | |
| }, | |
| { | |
| "epoch": 0.648902388513047, | |
| "grad_norm": 21.298200607299805, | |
| "learning_rate": 5.802998467231075e-08, | |
| "loss": 0.8913, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.6502830318928621, | |
| "grad_norm": 9.219822883605957, | |
| "learning_rate": 5.762449725789462e-08, | |
| "loss": 0.8855, | |
| "step": 4710 | |
| }, | |
| { | |
| "epoch": 0.6516636752726771, | |
| "grad_norm": 9.993160247802734, | |
| "learning_rate": 5.7219857307419117e-08, | |
| "loss": 0.8875, | |
| "step": 4720 | |
| }, | |
| { | |
| "epoch": 0.653044318652492, | |
| "grad_norm": 22.09342384338379, | |
| "learning_rate": 5.6816072913242906e-08, | |
| "loss": 0.8781, | |
| "step": 4730 | |
| }, | |
| { | |
| "epoch": 0.654424962032307, | |
| "grad_norm": 15.692350387573242, | |
| "learning_rate": 5.641315215061446e-08, | |
| "loss": 0.8773, | |
| "step": 4740 | |
| }, | |
| { | |
| "epoch": 0.6558056054121221, | |
| "grad_norm": 12.794647216796875, | |
| "learning_rate": 5.601110307751062e-08, | |
| "loss": 0.8634, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 0.657186248791937, | |
| "grad_norm": 10.979103088378906, | |
| "learning_rate": 5.5609933734475234e-08, | |
| "loss": 0.8775, | |
| "step": 4760 | |
| }, | |
| { | |
| "epoch": 0.658566892171752, | |
| "grad_norm": 11.532218933105469, | |
| "learning_rate": 5.5209652144458595e-08, | |
| "loss": 0.872, | |
| "step": 4770 | |
| }, | |
| { | |
| "epoch": 0.659947535551567, | |
| "grad_norm": 20.71472930908203, | |
| "learning_rate": 5.481026631265704e-08, | |
| "loss": 0.8804, | |
| "step": 4780 | |
| }, | |
| { | |
| "epoch": 0.6613281789313821, | |
| "grad_norm": 18.97435188293457, | |
| "learning_rate": 5.4411784226352444e-08, | |
| "loss": 0.8622, | |
| "step": 4790 | |
| }, | |
| { | |
| "epoch": 0.662708822311197, | |
| "grad_norm": 9.784866333007812, | |
| "learning_rate": 5.401421385475295e-08, | |
| "loss": 0.8878, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.664089465691012, | |
| "grad_norm": 12.122749328613281, | |
| "learning_rate": 5.3617563148833344e-08, | |
| "loss": 0.8585, | |
| "step": 4810 | |
| }, | |
| { | |
| "epoch": 0.665470109070827, | |
| "grad_norm": 13.080982208251953, | |
| "learning_rate": 5.322184004117607e-08, | |
| "loss": 0.8644, | |
| "step": 4820 | |
| }, | |
| { | |
| "epoch": 0.666850752450642, | |
| "grad_norm": 12.814577102661133, | |
| "learning_rate": 5.282705244581267e-08, | |
| "loss": 0.9015, | |
| "step": 4830 | |
| }, | |
| { | |
| "epoch": 0.668231395830457, | |
| "grad_norm": 9.391085624694824, | |
| "learning_rate": 5.243320825806547e-08, | |
| "loss": 0.9012, | |
| "step": 4840 | |
| }, | |
| { | |
| "epoch": 0.669612039210272, | |
| "grad_norm": 12.146528244018555, | |
| "learning_rate": 5.204031535438952e-08, | |
| "loss": 0.8929, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 0.6709926825900869, | |
| "grad_norm": 59.82147979736328, | |
| "learning_rate": 5.164838159221547e-08, | |
| "loss": 0.8948, | |
| "step": 4860 | |
| }, | |
| { | |
| "epoch": 0.672373325969902, | |
| "grad_norm": 10.037934303283691, | |
| "learning_rate": 5.125741480979195e-08, | |
| "loss": 0.8841, | |
| "step": 4870 | |
| }, | |
| { | |
| "epoch": 0.673753969349717, | |
| "grad_norm": 11.494805335998535, | |
| "learning_rate": 5.086742282602918e-08, | |
| "loss": 0.871, | |
| "step": 4880 | |
| }, | |
| { | |
| "epoch": 0.675134612729532, | |
| "grad_norm": 16.17136001586914, | |
| "learning_rate": 5.047841344034245e-08, | |
| "loss": 0.8901, | |
| "step": 4890 | |
| }, | |
| { | |
| "epoch": 0.6765152561093469, | |
| "grad_norm": 15.18946647644043, | |
| "learning_rate": 5.009039443249612e-08, | |
| "loss": 0.8893, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.677895899489162, | |
| "grad_norm": 9.07051944732666, | |
| "learning_rate": 4.97033735624481e-08, | |
| "loss": 0.8677, | |
| "step": 4910 | |
| }, | |
| { | |
| "epoch": 0.679276542868977, | |
| "grad_norm": 12.248629570007324, | |
| "learning_rate": 4.931735857019468e-08, | |
| "loss": 0.8827, | |
| "step": 4920 | |
| }, | |
| { | |
| "epoch": 0.6806571862487919, | |
| "grad_norm": 10.949799537658691, | |
| "learning_rate": 4.893235717561549e-08, | |
| "loss": 0.9134, | |
| "step": 4930 | |
| }, | |
| { | |
| "epoch": 0.6820378296286069, | |
| "grad_norm": 11.260576248168945, | |
| "learning_rate": 4.854837707831963e-08, | |
| "loss": 0.8728, | |
| "step": 4940 | |
| }, | |
| { | |
| "epoch": 0.683418473008422, | |
| "grad_norm": 9.589130401611328, | |
| "learning_rate": 4.8165425957491054e-08, | |
| "loss": 0.898, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 0.6847991163882369, | |
| "grad_norm": 15.198226928710938, | |
| "learning_rate": 4.7783511471735504e-08, | |
| "loss": 0.8753, | |
| "step": 4960 | |
| }, | |
| { | |
| "epoch": 0.6861797597680519, | |
| "grad_norm": 12.14223575592041, | |
| "learning_rate": 4.740264125892706e-08, | |
| "loss": 0.8861, | |
| "step": 4970 | |
| }, | |
| { | |
| "epoch": 0.6875604031478669, | |
| "grad_norm": 9.604902267456055, | |
| "learning_rate": 4.702282293605551e-08, | |
| "loss": 0.8867, | |
| "step": 4980 | |
| }, | |
| { | |
| "epoch": 0.688941046527682, | |
| "grad_norm": 11.847143173217773, | |
| "learning_rate": 4.664406409907399e-08, | |
| "loss": 0.8811, | |
| "step": 4990 | |
| }, | |
| { | |
| "epoch": 0.6903216899074969, | |
| "grad_norm": 10.780383110046387, | |
| "learning_rate": 4.626637232274709e-08, | |
| "loss": 0.8825, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.6917023332873119, | |
| "grad_norm": 17.666549682617188, | |
| "learning_rate": 4.588975516049926e-08, | |
| "loss": 0.9029, | |
| "step": 5010 | |
| }, | |
| { | |
| "epoch": 0.6930829766671269, | |
| "grad_norm": 14.401239395141602, | |
| "learning_rate": 4.5514220144263896e-08, | |
| "loss": 0.8791, | |
| "step": 5020 | |
| }, | |
| { | |
| "epoch": 0.6944636200469418, | |
| "grad_norm": 9.24218463897705, | |
| "learning_rate": 4.513977478433275e-08, | |
| "loss": 0.9064, | |
| "step": 5030 | |
| }, | |
| { | |
| "epoch": 0.6958442634267569, | |
| "grad_norm": 8.766172409057617, | |
| "learning_rate": 4.476642656920545e-08, | |
| "loss": 0.9007, | |
| "step": 5040 | |
| }, | |
| { | |
| "epoch": 0.6972249068065719, | |
| "grad_norm": 18.305450439453125, | |
| "learning_rate": 4.439418296544004e-08, | |
| "loss": 0.861, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 0.6986055501863868, | |
| "grad_norm": 10.024104118347168, | |
| "learning_rate": 4.402305141750351e-08, | |
| "loss": 0.85, | |
| "step": 5060 | |
| }, | |
| { | |
| "epoch": 0.6999861935662018, | |
| "grad_norm": 10.37547492980957, | |
| "learning_rate": 4.3653039347622947e-08, | |
| "loss": 0.8894, | |
| "step": 5070 | |
| }, | |
| { | |
| "epoch": 0.7013668369460169, | |
| "grad_norm": 9.432260513305664, | |
| "learning_rate": 4.328415415563711e-08, | |
| "loss": 0.8719, | |
| "step": 5080 | |
| }, | |
| { | |
| "epoch": 0.7027474803258318, | |
| "grad_norm": 10.539888381958008, | |
| "learning_rate": 4.291640321884835e-08, | |
| "loss": 0.879, | |
| "step": 5090 | |
| }, | |
| { | |
| "epoch": 0.7041281237056468, | |
| "grad_norm": 10.342575073242188, | |
| "learning_rate": 4.254979389187519e-08, | |
| "loss": 0.8845, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 0.7055087670854618, | |
| "grad_norm": 11.448867797851562, | |
| "learning_rate": 4.218433350650535e-08, | |
| "loss": 0.8519, | |
| "step": 5110 | |
| }, | |
| { | |
| "epoch": 0.7068894104652769, | |
| "grad_norm": 13.00378131866455, | |
| "learning_rate": 4.182002937154871e-08, | |
| "loss": 0.9049, | |
| "step": 5120 | |
| }, | |
| { | |
| "epoch": 0.7082700538450918, | |
| "grad_norm": 16.519426345825195, | |
| "learning_rate": 4.145688877269157e-08, | |
| "loss": 0.861, | |
| "step": 5130 | |
| }, | |
| { | |
| "epoch": 0.7096506972249068, | |
| "grad_norm": 11.879060745239258, | |
| "learning_rate": 4.10949189723507e-08, | |
| "loss": 0.8961, | |
| "step": 5140 | |
| }, | |
| { | |
| "epoch": 0.7110313406047218, | |
| "grad_norm": 15.536749839782715, | |
| "learning_rate": 4.073412720952822e-08, | |
| "loss": 0.8792, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 0.7124119839845368, | |
| "grad_norm": 12.497264862060547, | |
| "learning_rate": 4.0374520699666805e-08, | |
| "loss": 0.8961, | |
| "step": 5160 | |
| }, | |
| { | |
| "epoch": 0.7137926273643518, | |
| "grad_norm": 9.491691589355469, | |
| "learning_rate": 4.0016106634505234e-08, | |
| "loss": 0.8832, | |
| "step": 5170 | |
| }, | |
| { | |
| "epoch": 0.7151732707441668, | |
| "grad_norm": 9.699114799499512, | |
| "learning_rate": 3.965889218193478e-08, | |
| "loss": 0.8835, | |
| "step": 5180 | |
| }, | |
| { | |
| "epoch": 0.7165539141239817, | |
| "grad_norm": 14.406123161315918, | |
| "learning_rate": 3.9302884485855856e-08, | |
| "loss": 0.8639, | |
| "step": 5190 | |
| }, | |
| { | |
| "epoch": 0.7179345575037968, | |
| "grad_norm": 11.906084060668945, | |
| "learning_rate": 3.8948090666034885e-08, | |
| "loss": 0.88, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 0.7193152008836118, | |
| "grad_norm": 10.461540222167969, | |
| "learning_rate": 3.859451781796219e-08, | |
| "loss": 0.906, | |
| "step": 5210 | |
| }, | |
| { | |
| "epoch": 0.7206958442634268, | |
| "grad_norm": 9.694464683532715, | |
| "learning_rate": 3.8242173012709966e-08, | |
| "loss": 0.8617, | |
| "step": 5220 | |
| }, | |
| { | |
| "epoch": 0.7220764876432417, | |
| "grad_norm": 12.70252513885498, | |
| "learning_rate": 3.789106329679087e-08, | |
| "loss": 0.9004, | |
| "step": 5230 | |
| }, | |
| { | |
| "epoch": 0.7234571310230568, | |
| "grad_norm": 12.003303527832031, | |
| "learning_rate": 3.754119569201716e-08, | |
| "loss": 0.8769, | |
| "step": 5240 | |
| }, | |
| { | |
| "epoch": 0.7248377744028718, | |
| "grad_norm": 14.755000114440918, | |
| "learning_rate": 3.719257719536022e-08, | |
| "loss": 0.912, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 0.7262184177826867, | |
| "grad_norm": 15.1683988571167, | |
| "learning_rate": 3.68452147788105e-08, | |
| "loss": 0.8866, | |
| "step": 5260 | |
| }, | |
| { | |
| "epoch": 0.7275990611625017, | |
| "grad_norm": 12.824217796325684, | |
| "learning_rate": 3.6499115389238455e-08, | |
| "loss": 0.8882, | |
| "step": 5270 | |
| }, | |
| { | |
| "epoch": 0.7289797045423168, | |
| "grad_norm": 14.749849319458008, | |
| "learning_rate": 3.6154285948255134e-08, | |
| "loss": 0.8638, | |
| "step": 5280 | |
| }, | |
| { | |
| "epoch": 0.7303603479221317, | |
| "grad_norm": 9.494807243347168, | |
| "learning_rate": 3.581073335207416e-08, | |
| "loss": 0.8815, | |
| "step": 5290 | |
| }, | |
| { | |
| "epoch": 0.7317409913019467, | |
| "grad_norm": 21.002761840820312, | |
| "learning_rate": 3.5468464471373616e-08, | |
| "loss": 0.8864, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 0.7331216346817617, | |
| "grad_norm": 13.743839263916016, | |
| "learning_rate": 3.5127486151158545e-08, | |
| "loss": 0.8824, | |
| "step": 5310 | |
| }, | |
| { | |
| "epoch": 0.7345022780615766, | |
| "grad_norm": 10.033440589904785, | |
| "learning_rate": 3.4787805210624385e-08, | |
| "loss": 0.8668, | |
| "step": 5320 | |
| }, | |
| { | |
| "epoch": 0.7358829214413917, | |
| "grad_norm": 14.06644058227539, | |
| "learning_rate": 3.444942844302031e-08, | |
| "loss": 0.8726, | |
| "step": 5330 | |
| }, | |
| { | |
| "epoch": 0.7372635648212067, | |
| "grad_norm": 9.101497650146484, | |
| "learning_rate": 3.4112362615513357e-08, | |
| "loss": 0.8947, | |
| "step": 5340 | |
| }, | |
| { | |
| "epoch": 0.7386442082010217, | |
| "grad_norm": 13.168549537658691, | |
| "learning_rate": 3.3776614469053324e-08, | |
| "loss": 0.896, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 0.7400248515808366, | |
| "grad_norm": 10.807592391967773, | |
| "learning_rate": 3.344219071823776e-08, | |
| "loss": 0.8762, | |
| "step": 5360 | |
| }, | |
| { | |
| "epoch": 0.7414054949606517, | |
| "grad_norm": 16.737186431884766, | |
| "learning_rate": 3.310909805117775e-08, | |
| "loss": 0.8944, | |
| "step": 5370 | |
| }, | |
| { | |
| "epoch": 0.7427861383404667, | |
| "grad_norm": 12.22408676147461, | |
| "learning_rate": 3.277734312936419e-08, | |
| "loss": 0.8793, | |
| "step": 5380 | |
| }, | |
| { | |
| "epoch": 0.7441667817202816, | |
| "grad_norm": 11.596902847290039, | |
| "learning_rate": 3.244693258753439e-08, | |
| "loss": 0.8795, | |
| "step": 5390 | |
| }, | |
| { | |
| "epoch": 0.7455474251000966, | |
| "grad_norm": 9.242010116577148, | |
| "learning_rate": 3.2117873033539725e-08, | |
| "loss": 0.8967, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 0.7469280684799117, | |
| "grad_norm": 14.697638511657715, | |
| "learning_rate": 3.179017104821319e-08, | |
| "loss": 0.8698, | |
| "step": 5410 | |
| }, | |
| { | |
| "epoch": 0.7483087118597266, | |
| "grad_norm": 14.949028968811035, | |
| "learning_rate": 3.14638331852378e-08, | |
| "loss": 0.8843, | |
| "step": 5420 | |
| }, | |
| { | |
| "epoch": 0.7496893552395416, | |
| "grad_norm": 12.559296607971191, | |
| "learning_rate": 3.113886597101571e-08, | |
| "loss": 0.8805, | |
| "step": 5430 | |
| }, | |
| { | |
| "epoch": 0.7510699986193566, | |
| "grad_norm": 9.95313835144043, | |
| "learning_rate": 3.08152759045376e-08, | |
| "loss": 0.868, | |
| "step": 5440 | |
| }, | |
| { | |
| "epoch": 0.7524506419991717, | |
| "grad_norm": 10.255855560302734, | |
| "learning_rate": 3.049306945725263e-08, | |
| "loss": 0.9027, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 0.7538312853789866, | |
| "grad_norm": 14.607210159301758, | |
| "learning_rate": 3.017225307293911e-08, | |
| "loss": 0.8827, | |
| "step": 5460 | |
| }, | |
| { | |
| "epoch": 0.7552119287588016, | |
| "grad_norm": 11.993898391723633, | |
| "learning_rate": 2.985283316757563e-08, | |
| "loss": 0.9096, | |
| "step": 5470 | |
| }, | |
| { | |
| "epoch": 0.7565925721386166, | |
| "grad_norm": 10.91257381439209, | |
| "learning_rate": 2.953481612921267e-08, | |
| "loss": 0.8881, | |
| "step": 5480 | |
| }, | |
| { | |
| "epoch": 0.7579732155184316, | |
| "grad_norm": 10.943377494812012, | |
| "learning_rate": 2.9218208317844972e-08, | |
| "loss": 0.8949, | |
| "step": 5490 | |
| }, | |
| { | |
| "epoch": 0.7593538588982466, | |
| "grad_norm": 12.161398887634277, | |
| "learning_rate": 2.8903016065284137e-08, | |
| "loss": 0.9184, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.7607345022780616, | |
| "grad_norm": 8.598383903503418, | |
| "learning_rate": 2.8589245675032237e-08, | |
| "loss": 0.888, | |
| "step": 5510 | |
| }, | |
| { | |
| "epoch": 0.7621151456578765, | |
| "grad_norm": 9.565353393554688, | |
| "learning_rate": 2.8276903422155706e-08, | |
| "loss": 0.8587, | |
| "step": 5520 | |
| }, | |
| { | |
| "epoch": 0.7634957890376916, | |
| "grad_norm": 12.679278373718262, | |
| "learning_rate": 2.7965995553159616e-08, | |
| "loss": 0.8879, | |
| "step": 5530 | |
| }, | |
| { | |
| "epoch": 0.7648764324175066, | |
| "grad_norm": 10.56402587890625, | |
| "learning_rate": 2.7656528285863034e-08, | |
| "loss": 0.8956, | |
| "step": 5540 | |
| }, | |
| { | |
| "epoch": 0.7662570757973216, | |
| "grad_norm": 13.341521263122559, | |
| "learning_rate": 2.7348507809274535e-08, | |
| "loss": 0.8942, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 0.7676377191771365, | |
| "grad_norm": 10.266841888427734, | |
| "learning_rate": 2.7041940283468445e-08, | |
| "loss": 0.9031, | |
| "step": 5560 | |
| }, | |
| { | |
| "epoch": 0.7690183625569516, | |
| "grad_norm": 8.802313804626465, | |
| "learning_rate": 2.6736831839461716e-08, | |
| "loss": 0.8624, | |
| "step": 5570 | |
| }, | |
| { | |
| "epoch": 0.7703990059367666, | |
| "grad_norm": 9.554488182067871, | |
| "learning_rate": 2.6433188579091127e-08, | |
| "loss": 0.8839, | |
| "step": 5580 | |
| }, | |
| { | |
| "epoch": 0.7717796493165815, | |
| "grad_norm": 10.517236709594727, | |
| "learning_rate": 2.6131016574891475e-08, | |
| "loss": 0.8686, | |
| "step": 5590 | |
| }, | |
| { | |
| "epoch": 0.7731602926963965, | |
| "grad_norm": 12.35927963256836, | |
| "learning_rate": 2.5830321869974024e-08, | |
| "loss": 0.8899, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 0.7745409360762116, | |
| "grad_norm": 14.404338836669922, | |
| "learning_rate": 2.5531110477905648e-08, | |
| "loss": 0.9076, | |
| "step": 5610 | |
| }, | |
| { | |
| "epoch": 0.7759215794560265, | |
| "grad_norm": 12.745739936828613, | |
| "learning_rate": 2.5233388382588582e-08, | |
| "loss": 0.8723, | |
| "step": 5620 | |
| }, | |
| { | |
| "epoch": 0.7773022228358415, | |
| "grad_norm": 16.569324493408203, | |
| "learning_rate": 2.493716153814076e-08, | |
| "loss": 0.8764, | |
| "step": 5630 | |
| }, | |
| { | |
| "epoch": 0.7786828662156565, | |
| "grad_norm": 16.94318199157715, | |
| "learning_rate": 2.4642435868776623e-08, | |
| "loss": 0.8823, | |
| "step": 5640 | |
| }, | |
| { | |
| "epoch": 0.7800635095954714, | |
| "grad_norm": 10.555295944213867, | |
| "learning_rate": 2.4349217268688927e-08, | |
| "loss": 0.8814, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 0.7814441529752865, | |
| "grad_norm": 10.818206787109375, | |
| "learning_rate": 2.405751160193057e-08, | |
| "loss": 0.8893, | |
| "step": 5660 | |
| }, | |
| { | |
| "epoch": 0.7828247963551015, | |
| "grad_norm": 19.33257484436035, | |
| "learning_rate": 2.3767324702297398e-08, | |
| "loss": 0.8503, | |
| "step": 5670 | |
| }, | |
| { | |
| "epoch": 0.7842054397349165, | |
| "grad_norm": 11.430990219116211, | |
| "learning_rate": 2.3478662373211643e-08, | |
| "loss": 0.8958, | |
| "step": 5680 | |
| }, | |
| { | |
| "epoch": 0.7855860831147314, | |
| "grad_norm": 10.336344718933105, | |
| "learning_rate": 2.319153038760575e-08, | |
| "loss": 0.8925, | |
| "step": 5690 | |
| }, | |
| { | |
| "epoch": 0.7869667264945465, | |
| "grad_norm": 9.480130195617676, | |
| "learning_rate": 2.2905934487806965e-08, | |
| "loss": 0.8692, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 0.7883473698743615, | |
| "grad_norm": 9.794612884521484, | |
| "learning_rate": 2.262188038542253e-08, | |
| "loss": 0.883, | |
| "step": 5710 | |
| }, | |
| { | |
| "epoch": 0.7897280132541764, | |
| "grad_norm": 19.427499771118164, | |
| "learning_rate": 2.233937376122529e-08, | |
| "loss": 0.8805, | |
| "step": 5720 | |
| }, | |
| { | |
| "epoch": 0.7911086566339914, | |
| "grad_norm": 12.043346405029297, | |
| "learning_rate": 2.2058420265040368e-08, | |
| "loss": 0.9002, | |
| "step": 5730 | |
| }, | |
| { | |
| "epoch": 0.7924893000138065, | |
| "grad_norm": 11.101466178894043, | |
| "learning_rate": 2.1779025515631956e-08, | |
| "loss": 0.8886, | |
| "step": 5740 | |
| }, | |
| { | |
| "epoch": 0.7938699433936214, | |
| "grad_norm": 13.886466026306152, | |
| "learning_rate": 2.150119510059094e-08, | |
| "loss": 0.8822, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 0.7952505867734364, | |
| "grad_norm": 10.427753448486328, | |
| "learning_rate": 2.1224934576223308e-08, | |
| "loss": 0.8486, | |
| "step": 5760 | |
| }, | |
| { | |
| "epoch": 0.7966312301532514, | |
| "grad_norm": 14.40750789642334, | |
| "learning_rate": 2.095024946743892e-08, | |
| "loss": 0.8803, | |
| "step": 5770 | |
| }, | |
| { | |
| "epoch": 0.7980118735330665, | |
| "grad_norm": 11.908204078674316, | |
| "learning_rate": 2.0677145267641004e-08, | |
| "loss": 0.8823, | |
| "step": 5780 | |
| }, | |
| { | |
| "epoch": 0.7993925169128814, | |
| "grad_norm": 17.171079635620117, | |
| "learning_rate": 2.0405627438616425e-08, | |
| "loss": 0.8523, | |
| "step": 5790 | |
| }, | |
| { | |
| "epoch": 0.8007731602926964, | |
| "grad_norm": 9.826622009277344, | |
| "learning_rate": 2.0135701410426165e-08, | |
| "loss": 0.8885, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 0.8021538036725114, | |
| "grad_norm": 14.045727729797363, | |
| "learning_rate": 1.986737258129716e-08, | |
| "loss": 0.8716, | |
| "step": 5810 | |
| }, | |
| { | |
| "epoch": 0.8035344470523264, | |
| "grad_norm": 8.29520034790039, | |
| "learning_rate": 1.960064631751398e-08, | |
| "loss": 0.8707, | |
| "step": 5820 | |
| }, | |
| { | |
| "epoch": 0.8049150904321414, | |
| "grad_norm": 16.575197219848633, | |
| "learning_rate": 1.933552795331158e-08, | |
| "loss": 0.8854, | |
| "step": 5830 | |
| }, | |
| { | |
| "epoch": 0.8062957338119564, | |
| "grad_norm": 16.817964553833008, | |
| "learning_rate": 1.907202279076877e-08, | |
| "loss": 0.8844, | |
| "step": 5840 | |
| }, | |
| { | |
| "epoch": 0.8076763771917713, | |
| "grad_norm": 14.492928504943848, | |
| "learning_rate": 1.8810136099702046e-08, | |
| "loss": 0.8715, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 0.8090570205715863, | |
| "grad_norm": 11.814017295837402, | |
| "learning_rate": 1.8549873117560232e-08, | |
| "loss": 0.8935, | |
| "step": 5860 | |
| }, | |
| { | |
| "epoch": 0.8104376639514014, | |
| "grad_norm": 13.838473320007324, | |
| "learning_rate": 1.8291239049319762e-08, | |
| "loss": 0.8726, | |
| "step": 5870 | |
| }, | |
| { | |
| "epoch": 0.8118183073312163, | |
| "grad_norm": 13.624913215637207, | |
| "learning_rate": 1.8034239067380552e-08, | |
| "loss": 0.8894, | |
| "step": 5880 | |
| }, | |
| { | |
| "epoch": 0.8131989507110313, | |
| "grad_norm": 9.358819961547852, | |
| "learning_rate": 1.77788783114625e-08, | |
| "loss": 0.8642, | |
| "step": 5890 | |
| }, | |
| { | |
| "epoch": 0.8145795940908463, | |
| "grad_norm": 14.413737297058105, | |
| "learning_rate": 1.7525161888502927e-08, | |
| "loss": 0.868, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 0.8159602374706614, | |
| "grad_norm": 11.16862964630127, | |
| "learning_rate": 1.7273094872554128e-08, | |
| "loss": 0.8682, | |
| "step": 5910 | |
| }, | |
| { | |
| "epoch": 0.8173408808504763, | |
| "grad_norm": 11.230937004089355, | |
| "learning_rate": 1.7022682304682136e-08, | |
| "loss": 0.8897, | |
| "step": 5920 | |
| }, | |
| { | |
| "epoch": 0.8187215242302913, | |
| "grad_norm": 13.970176696777344, | |
| "learning_rate": 1.6773929192865798e-08, | |
| "loss": 0.8815, | |
| "step": 5930 | |
| }, | |
| { | |
| "epoch": 0.8201021676101063, | |
| "grad_norm": 9.744919776916504, | |
| "learning_rate": 1.6526840511896657e-08, | |
| "loss": 0.8784, | |
| "step": 5940 | |
| }, | |
| { | |
| "epoch": 0.8214828109899213, | |
| "grad_norm": 12.269463539123535, | |
| "learning_rate": 1.6281421203279434e-08, | |
| "loss": 0.8946, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 0.8228634543697363, | |
| "grad_norm": 13.358258247375488, | |
| "learning_rate": 1.6037676175133264e-08, | |
| "loss": 0.8839, | |
| "step": 5960 | |
| }, | |
| { | |
| "epoch": 0.8242440977495513, | |
| "grad_norm": 14.507047653198242, | |
| "learning_rate": 1.5795610302093377e-08, | |
| "loss": 0.8703, | |
| "step": 5970 | |
| }, | |
| { | |
| "epoch": 0.8256247411293662, | |
| "grad_norm": 17.883962631225586, | |
| "learning_rate": 1.5555228425213907e-08, | |
| "loss": 0.8835, | |
| "step": 5980 | |
| }, | |
| { | |
| "epoch": 0.8270053845091813, | |
| "grad_norm": 9.623534202575684, | |
| "learning_rate": 1.5316535351870752e-08, | |
| "loss": 0.8861, | |
| "step": 5990 | |
| }, | |
| { | |
| "epoch": 0.8283860278889963, | |
| "grad_norm": 11.326260566711426, | |
| "learning_rate": 1.5079535855665613e-08, | |
| "loss": 0.8706, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.8297666712688113, | |
| "grad_norm": 10.933159828186035, | |
| "learning_rate": 1.4844234676330547e-08, | |
| "loss": 0.8793, | |
| "step": 6010 | |
| }, | |
| { | |
| "epoch": 0.8311473146486262, | |
| "grad_norm": 10.065755844116211, | |
| "learning_rate": 1.4610636519633047e-08, | |
| "loss": 0.8889, | |
| "step": 6020 | |
| }, | |
| { | |
| "epoch": 0.8325279580284413, | |
| "grad_norm": 10.439647674560547, | |
| "learning_rate": 1.4378746057282043e-08, | |
| "loss": 0.8793, | |
| "step": 6030 | |
| }, | |
| { | |
| "epoch": 0.8339086014082563, | |
| "grad_norm": 12.923277854919434, | |
| "learning_rate": 1.4148567926834465e-08, | |
| "loss": 0.8956, | |
| "step": 6040 | |
| }, | |
| { | |
| "epoch": 0.8352892447880712, | |
| "grad_norm": 18.733732223510742, | |
| "learning_rate": 1.3920106731602365e-08, | |
| "loss": 0.8696, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 0.8366698881678862, | |
| "grad_norm": 14.746212005615234, | |
| "learning_rate": 1.369336704056111e-08, | |
| "loss": 0.8583, | |
| "step": 6060 | |
| }, | |
| { | |
| "epoch": 0.8380505315477013, | |
| "grad_norm": 12.682589530944824, | |
| "learning_rate": 1.3468353388257725e-08, | |
| "loss": 0.8615, | |
| "step": 6070 | |
| }, | |
| { | |
| "epoch": 0.8394311749275162, | |
| "grad_norm": 13.228649139404297, | |
| "learning_rate": 1.3245070274720416e-08, | |
| "loss": 0.889, | |
| "step": 6080 | |
| }, | |
| { | |
| "epoch": 0.8408118183073312, | |
| "grad_norm": 9.542222023010254, | |
| "learning_rate": 1.3023522165368439e-08, | |
| "loss": 0.8744, | |
| "step": 6090 | |
| }, | |
| { | |
| "epoch": 0.8421924616871462, | |
| "grad_norm": 14.271943092346191, | |
| "learning_rate": 1.2803713490922918e-08, | |
| "loss": 0.8883, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 0.8435731050669611, | |
| "grad_norm": 17.61199188232422, | |
| "learning_rate": 1.2585648647318126e-08, | |
| "loss": 0.8735, | |
| "step": 6110 | |
| }, | |
| { | |
| "epoch": 0.8449537484467762, | |
| "grad_norm": 14.385493278503418, | |
| "learning_rate": 1.2369331995613663e-08, | |
| "loss": 0.8924, | |
| "step": 6120 | |
| }, | |
| { | |
| "epoch": 0.8463343918265912, | |
| "grad_norm": 16.303983688354492, | |
| "learning_rate": 1.2154767861907078e-08, | |
| "loss": 0.9089, | |
| "step": 6130 | |
| }, | |
| { | |
| "epoch": 0.8477150352064062, | |
| "grad_norm": 8.969123840332031, | |
| "learning_rate": 1.1941960537247631e-08, | |
| "loss": 0.8932, | |
| "step": 6140 | |
| }, | |
| { | |
| "epoch": 0.8490956785862211, | |
| "grad_norm": 12.357159614562988, | |
| "learning_rate": 1.1730914277550219e-08, | |
| "loss": 0.8984, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 0.8504763219660362, | |
| "grad_norm": 11.281810760498047, | |
| "learning_rate": 1.1521633303510347e-08, | |
| "loss": 0.8699, | |
| "step": 6160 | |
| }, | |
| { | |
| "epoch": 0.8518569653458512, | |
| "grad_norm": 9.341175079345703, | |
| "learning_rate": 1.131412180051975e-08, | |
| "loss": 0.8711, | |
| "step": 6170 | |
| }, | |
| { | |
| "epoch": 0.8532376087256661, | |
| "grad_norm": 13.833061218261719, | |
| "learning_rate": 1.1108383918582676e-08, | |
| "loss": 0.8836, | |
| "step": 6180 | |
| }, | |
| { | |
| "epoch": 0.8546182521054811, | |
| "grad_norm": 11.766849517822266, | |
| "learning_rate": 1.0904423772232884e-08, | |
| "loss": 0.8739, | |
| "step": 6190 | |
| }, | |
| { | |
| "epoch": 0.8559988954852962, | |
| "grad_norm": 8.81930160522461, | |
| "learning_rate": 1.0702245440451351e-08, | |
| "loss": 0.897, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 0.8573795388651111, | |
| "grad_norm": 13.813037872314453, | |
| "learning_rate": 1.0501852966584667e-08, | |
| "loss": 0.8717, | |
| "step": 6210 | |
| }, | |
| { | |
| "epoch": 0.8587601822449261, | |
| "grad_norm": 9.942933082580566, | |
| "learning_rate": 1.0303250358264227e-08, | |
| "loss": 0.894, | |
| "step": 6220 | |
| }, | |
| { | |
| "epoch": 0.8601408256247411, | |
| "grad_norm": 12.416991233825684, | |
| "learning_rate": 1.010644158732613e-08, | |
| "loss": 0.8781, | |
| "step": 6230 | |
| }, | |
| { | |
| "epoch": 0.8615214690045562, | |
| "grad_norm": 15.107084274291992, | |
| "learning_rate": 9.911430589731584e-09, | |
| "loss": 0.8879, | |
| "step": 6240 | |
| }, | |
| { | |
| "epoch": 0.8629021123843711, | |
| "grad_norm": 18.36236000061035, | |
| "learning_rate": 9.718221265488325e-09, | |
| "loss": 0.8712, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 0.8642827557641861, | |
| "grad_norm": 14.529742240905762, | |
| "learning_rate": 9.52681747857259e-09, | |
| "loss": 0.882, | |
| "step": 6260 | |
| }, | |
| { | |
| "epoch": 0.8656633991440011, | |
| "grad_norm": 13.656230926513672, | |
| "learning_rate": 9.337223056851817e-09, | |
| "loss": 0.8863, | |
| "step": 6270 | |
| }, | |
| { | |
| "epoch": 0.8670440425238161, | |
| "grad_norm": 8.240508079528809, | |
| "learning_rate": 9.14944179200815e-09, | |
| "loss": 0.8762, | |
| "step": 6280 | |
| }, | |
| { | |
| "epoch": 0.8684246859036311, | |
| "grad_norm": 19.294933319091797, | |
| "learning_rate": 8.963477439462486e-09, | |
| "loss": 0.8831, | |
| "step": 6290 | |
| }, | |
| { | |
| "epoch": 0.8698053292834461, | |
| "grad_norm": 8.610947608947754, | |
| "learning_rate": 8.779333718299519e-09, | |
| "loss": 0.8671, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 0.871185972663261, | |
| "grad_norm": 11.317117691040039, | |
| "learning_rate": 8.597014311193317e-09, | |
| "loss": 0.8889, | |
| "step": 6310 | |
| }, | |
| { | |
| "epoch": 0.8725666160430761, | |
| "grad_norm": 12.359908103942871, | |
| "learning_rate": 8.416522864333587e-09, | |
| "loss": 0.8609, | |
| "step": 6320 | |
| }, | |
| { | |
| "epoch": 0.8739472594228911, | |
| "grad_norm": 9.896758079528809, | |
| "learning_rate": 8.23786298735285e-09, | |
| "loss": 0.8842, | |
| "step": 6330 | |
| }, | |
| { | |
| "epoch": 0.8753279028027061, | |
| "grad_norm": 11.469202041625977, | |
| "learning_rate": 8.061038253254238e-09, | |
| "loss": 0.8834, | |
| "step": 6340 | |
| }, | |
| { | |
| "epoch": 0.876708546182521, | |
| "grad_norm": 13.681952476501465, | |
| "learning_rate": 7.886052198339988e-09, | |
| "loss": 0.8705, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 0.8780891895623361, | |
| "grad_norm": 13.58003044128418, | |
| "learning_rate": 7.712908322140788e-09, | |
| "loss": 0.8859, | |
| "step": 6360 | |
| }, | |
| { | |
| "epoch": 0.8794698329421511, | |
| "grad_norm": 10.038475036621094, | |
| "learning_rate": 7.541610087345762e-09, | |
| "loss": 0.8602, | |
| "step": 6370 | |
| }, | |
| { | |
| "epoch": 0.880850476321966, | |
| "grad_norm": 10.786853790283203, | |
| "learning_rate": 7.372160919733128e-09, | |
| "loss": 0.8837, | |
| "step": 6380 | |
| }, | |
| { | |
| "epoch": 0.882231119701781, | |
| "grad_norm": 8.505433082580566, | |
| "learning_rate": 7.2045642081019155e-09, | |
| "loss": 0.8787, | |
| "step": 6390 | |
| }, | |
| { | |
| "epoch": 0.8836117630815961, | |
| "grad_norm": 10.103442192077637, | |
| "learning_rate": 7.038823304203922e-09, | |
| "loss": 0.8989, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 0.884992406461411, | |
| "grad_norm": 9.180789947509766, | |
| "learning_rate": 6.87494152267688e-09, | |
| "loss": 0.859, | |
| "step": 6410 | |
| }, | |
| { | |
| "epoch": 0.886373049841226, | |
| "grad_norm": 22.57208824157715, | |
| "learning_rate": 6.712922140978084e-09, | |
| "loss": 0.8708, | |
| "step": 6420 | |
| }, | |
| { | |
| "epoch": 0.887753693221041, | |
| "grad_norm": 9.427815437316895, | |
| "learning_rate": 6.552768399318842e-09, | |
| "loss": 0.8877, | |
| "step": 6430 | |
| }, | |
| { | |
| "epoch": 0.889134336600856, | |
| "grad_norm": 10.753376007080078, | |
| "learning_rate": 6.394483500599712e-09, | |
| "loss": 0.8858, | |
| "step": 6440 | |
| }, | |
| { | |
| "epoch": 0.890514979980671, | |
| "grad_norm": 9.80281925201416, | |
| "learning_rate": 6.238070610346435e-09, | |
| "loss": 0.8944, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 0.891895623360486, | |
| "grad_norm": 10.95217227935791, | |
| "learning_rate": 6.0835328566465675e-09, | |
| "loss": 0.8955, | |
| "step": 6460 | |
| }, | |
| { | |
| "epoch": 0.893276266740301, | |
| "grad_norm": 8.655145645141602, | |
| "learning_rate": 5.930873330087016e-09, | |
| "loss": 0.869, | |
| "step": 6470 | |
| }, | |
| { | |
| "epoch": 0.894656910120116, | |
| "grad_norm": 12.709726333618164, | |
| "learning_rate": 5.7800950836922045e-09, | |
| "loss": 0.8624, | |
| "step": 6480 | |
| }, | |
| { | |
| "epoch": 0.896037553499931, | |
| "grad_norm": 11.662334442138672, | |
| "learning_rate": 5.631201132862973e-09, | |
| "loss": 0.8934, | |
| "step": 6490 | |
| }, | |
| { | |
| "epoch": 0.897418196879746, | |
| "grad_norm": 10.503067970275879, | |
| "learning_rate": 5.484194455316304e-09, | |
| "loss": 0.887, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.8987988402595609, | |
| "grad_norm": 10.608843803405762, | |
| "learning_rate": 5.3390779910257644e-09, | |
| "loss": 0.8698, | |
| "step": 6510 | |
| }, | |
| { | |
| "epoch": 0.9001794836393759, | |
| "grad_norm": 10.961039543151855, | |
| "learning_rate": 5.195854642162734e-09, | |
| "loss": 0.8648, | |
| "step": 6520 | |
| }, | |
| { | |
| "epoch": 0.901560127019191, | |
| "grad_norm": 10.502799987792969, | |
| "learning_rate": 5.054527273038345e-09, | |
| "loss": 0.8712, | |
| "step": 6530 | |
| }, | |
| { | |
| "epoch": 0.9029407703990059, | |
| "grad_norm": 11.3438081741333, | |
| "learning_rate": 4.91509871004614e-09, | |
| "loss": 0.886, | |
| "step": 6540 | |
| }, | |
| { | |
| "epoch": 0.9043214137788209, | |
| "grad_norm": 8.995625495910645, | |
| "learning_rate": 4.777571741605657e-09, | |
| "loss": 0.8806, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 0.9057020571586359, | |
| "grad_norm": 7.603050231933594, | |
| "learning_rate": 4.6419491181066476e-09, | |
| "loss": 0.8504, | |
| "step": 6560 | |
| }, | |
| { | |
| "epoch": 0.907082700538451, | |
| "grad_norm": 9.927618980407715, | |
| "learning_rate": 4.508233551853957e-09, | |
| "loss": 0.8882, | |
| "step": 6570 | |
| }, | |
| { | |
| "epoch": 0.9084633439182659, | |
| "grad_norm": 14.684597969055176, | |
| "learning_rate": 4.3764277170133975e-09, | |
| "loss": 0.8816, | |
| "step": 6580 | |
| }, | |
| { | |
| "epoch": 0.9098439872980809, | |
| "grad_norm": 12.118766784667969, | |
| "learning_rate": 4.246534249558242e-09, | |
| "loss": 0.8861, | |
| "step": 6590 | |
| }, | |
| { | |
| "epoch": 0.9112246306778959, | |
| "grad_norm": 9.813715934753418, | |
| "learning_rate": 4.11855574721649e-09, | |
| "loss": 0.8498, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 0.9126052740577109, | |
| "grad_norm": 10.439380645751953, | |
| "learning_rate": 3.992494769418919e-09, | |
| "loss": 0.8774, | |
| "step": 6610 | |
| }, | |
| { | |
| "epoch": 0.9139859174375259, | |
| "grad_norm": 16.602672576904297, | |
| "learning_rate": 3.868353837247873e-09, | |
| "loss": 0.8698, | |
| "step": 6620 | |
| }, | |
| { | |
| "epoch": 0.9153665608173409, | |
| "grad_norm": 18.53183937072754, | |
| "learning_rate": 3.746135433386921e-09, | |
| "loss": 0.8965, | |
| "step": 6630 | |
| }, | |
| { | |
| "epoch": 0.9167472041971558, | |
| "grad_norm": 14.567218780517578, | |
| "learning_rate": 3.6258420020711554e-09, | |
| "loss": 0.8793, | |
| "step": 6640 | |
| }, | |
| { | |
| "epoch": 0.9181278475769709, | |
| "grad_norm": 11.848518371582031, | |
| "learning_rate": 3.5074759490382742e-09, | |
| "loss": 0.904, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 0.9195084909567859, | |
| "grad_norm": 16.070884704589844, | |
| "learning_rate": 3.391039641480531e-09, | |
| "loss": 0.8852, | |
| "step": 6660 | |
| }, | |
| { | |
| "epoch": 0.9208891343366008, | |
| "grad_norm": 16.751605987548828, | |
| "learning_rate": 3.276535407997383e-09, | |
| "loss": 0.8873, | |
| "step": 6670 | |
| }, | |
| { | |
| "epoch": 0.9222697777164158, | |
| "grad_norm": 10.462998390197754, | |
| "learning_rate": 3.163965538548874e-09, | |
| "loss": 0.8587, | |
| "step": 6680 | |
| }, | |
| { | |
| "epoch": 0.9236504210962309, | |
| "grad_norm": 11.892049789428711, | |
| "learning_rate": 3.0533322844099042e-09, | |
| "loss": 0.8735, | |
| "step": 6690 | |
| }, | |
| { | |
| "epoch": 0.9250310644760459, | |
| "grad_norm": 13.557024002075195, | |
| "learning_rate": 2.944637858125154e-09, | |
| "loss": 0.8915, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 0.9264117078558608, | |
| "grad_norm": 14.010783195495605, | |
| "learning_rate": 2.8378844334648654e-09, | |
| "loss": 0.8792, | |
| "step": 6710 | |
| }, | |
| { | |
| "epoch": 0.9277923512356758, | |
| "grad_norm": 12.493358612060547, | |
| "learning_rate": 2.7330741453813755e-09, | |
| "loss": 0.8683, | |
| "step": 6720 | |
| }, | |
| { | |
| "epoch": 0.9291729946154909, | |
| "grad_norm": 16.112028121948242, | |
| "learning_rate": 2.630209089966362e-09, | |
| "loss": 0.929, | |
| "step": 6730 | |
| }, | |
| { | |
| "epoch": 0.9305536379953058, | |
| "grad_norm": 16.063701629638672, | |
| "learning_rate": 2.529291324409011e-09, | |
| "loss": 0.8887, | |
| "step": 6740 | |
| }, | |
| { | |
| "epoch": 0.9319342813751208, | |
| "grad_norm": 18.000648498535156, | |
| "learning_rate": 2.4303228669548038e-09, | |
| "loss": 0.8668, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 0.9333149247549358, | |
| "grad_norm": 11.203133583068848, | |
| "learning_rate": 2.3333056968651844e-09, | |
| "loss": 0.8894, | |
| "step": 6760 | |
| }, | |
| { | |
| "epoch": 0.9346955681347507, | |
| "grad_norm": 16.775104522705078, | |
| "learning_rate": 2.2382417543780007e-09, | |
| "loss": 0.8828, | |
| "step": 6770 | |
| }, | |
| { | |
| "epoch": 0.9360762115145658, | |
| "grad_norm": 15.63203239440918, | |
| "learning_rate": 2.145132940668637e-09, | |
| "loss": 0.8747, | |
| "step": 6780 | |
| }, | |
| { | |
| "epoch": 0.9374568548943808, | |
| "grad_norm": 11.378479957580566, | |
| "learning_rate": 2.0539811178120316e-09, | |
| "loss": 0.8811, | |
| "step": 6790 | |
| }, | |
| { | |
| "epoch": 0.9388374982741958, | |
| "grad_norm": 14.024568557739258, | |
| "learning_rate": 1.964788108745463e-09, | |
| "loss": 0.8941, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 0.9402181416540107, | |
| "grad_norm": 13.124571800231934, | |
| "learning_rate": 1.877555697232047e-09, | |
| "loss": 0.886, | |
| "step": 6810 | |
| }, | |
| { | |
| "epoch": 0.9415987850338258, | |
| "grad_norm": 11.059711456298828, | |
| "learning_rate": 1.7922856278250853e-09, | |
| "loss": 0.8811, | |
| "step": 6820 | |
| }, | |
| { | |
| "epoch": 0.9429794284136408, | |
| "grad_norm": 17.20572853088379, | |
| "learning_rate": 1.7089796058331829e-09, | |
| "loss": 0.8683, | |
| "step": 6830 | |
| }, | |
| { | |
| "epoch": 0.9443600717934557, | |
| "grad_norm": 9.522912979125977, | |
| "learning_rate": 1.6276392972860876e-09, | |
| "loss": 0.8758, | |
| "step": 6840 | |
| }, | |
| { | |
| "epoch": 0.9457407151732707, | |
| "grad_norm": 8.982587814331055, | |
| "learning_rate": 1.548266328901493e-09, | |
| "loss": 0.8717, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 0.9471213585530858, | |
| "grad_norm": 11.5287446975708, | |
| "learning_rate": 1.470862288052388e-09, | |
| "loss": 0.8622, | |
| "step": 6860 | |
| }, | |
| { | |
| "epoch": 0.9485020019329007, | |
| "grad_norm": 25.898061752319336, | |
| "learning_rate": 1.3954287227353257e-09, | |
| "loss": 0.887, | |
| "step": 6870 | |
| }, | |
| { | |
| "epoch": 0.9498826453127157, | |
| "grad_norm": 12.208502769470215, | |
| "learning_rate": 1.3219671415395484e-09, | |
| "loss": 0.876, | |
| "step": 6880 | |
| }, | |
| { | |
| "epoch": 0.9512632886925307, | |
| "grad_norm": 9.247186660766602, | |
| "learning_rate": 1.2504790136167343e-09, | |
| "loss": 0.888, | |
| "step": 6890 | |
| }, | |
| { | |
| "epoch": 0.9526439320723458, | |
| "grad_norm": 14.410557746887207, | |
| "learning_rate": 1.1809657686516427e-09, | |
| "loss": 0.8705, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 0.9540245754521607, | |
| "grad_norm": 9.14966869354248, | |
| "learning_rate": 1.1134287968335377e-09, | |
| "loss": 0.8595, | |
| "step": 6910 | |
| }, | |
| { | |
| "epoch": 0.9554052188319757, | |
| "grad_norm": 10.05492877960205, | |
| "learning_rate": 1.0478694488283312e-09, | |
| "loss": 0.8851, | |
| "step": 6920 | |
| }, | |
| { | |
| "epoch": 0.9567858622117907, | |
| "grad_norm": 10.730325698852539, | |
| "learning_rate": 9.84289035751662e-10, | |
| "loss": 0.866, | |
| "step": 6930 | |
| }, | |
| { | |
| "epoch": 0.9581665055916057, | |
| "grad_norm": 27.26226043701172, | |
| "learning_rate": 9.226888291426038e-10, | |
| "loss": 0.894, | |
| "step": 6940 | |
| }, | |
| { | |
| "epoch": 0.9595471489714207, | |
| "grad_norm": 9.088162422180176, | |
| "learning_rate": 8.630700609382536e-10, | |
| "loss": 0.8742, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 0.9609277923512357, | |
| "grad_norm": 13.799763679504395, | |
| "learning_rate": 8.054339234490725e-10, | |
| "loss": 0.8791, | |
| "step": 6960 | |
| }, | |
| { | |
| "epoch": 0.9623084357310506, | |
| "grad_norm": 14.225488662719727, | |
| "learning_rate": 7.497815693351172e-10, | |
| "loss": 0.8826, | |
| "step": 6970 | |
| }, | |
| { | |
| "epoch": 0.9636890791108657, | |
| "grad_norm": 12.780132293701172, | |
| "learning_rate": 6.961141115829017e-10, | |
| "loss": 0.8805, | |
| "step": 6980 | |
| }, | |
| { | |
| "epoch": 0.9650697224906807, | |
| "grad_norm": 11.835436820983887, | |
| "learning_rate": 6.444326234831709e-10, | |
| "loss": 0.8991, | |
| "step": 6990 | |
| }, | |
| { | |
| "epoch": 0.9664503658704956, | |
| "grad_norm": 10.776239395141602, | |
| "learning_rate": 5.947381386094518e-10, | |
| "loss": 0.8476, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.9678310092503106, | |
| "grad_norm": 10.21164608001709, | |
| "learning_rate": 5.470316507973582e-10, | |
| "loss": 0.8718, | |
| "step": 7010 | |
| }, | |
| { | |
| "epoch": 0.9692116526301257, | |
| "grad_norm": 8.981840133666992, | |
| "learning_rate": 5.013141141247402e-10, | |
| "loss": 0.8618, | |
| "step": 7020 | |
| }, | |
| { | |
| "epoch": 0.9705922960099407, | |
| "grad_norm": 12.117026329040527, | |
| "learning_rate": 4.5758644289256574e-10, | |
| "loss": 0.9069, | |
| "step": 7030 | |
| }, | |
| { | |
| "epoch": 0.9719729393897556, | |
| "grad_norm": 11.438433647155762, | |
| "learning_rate": 4.158495116066807e-10, | |
| "loss": 0.8789, | |
| "step": 7040 | |
| }, | |
| { | |
| "epoch": 0.9733535827695706, | |
| "grad_norm": 14.741561889648438, | |
| "learning_rate": 3.761041549602884e-10, | |
| "loss": 0.9055, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 0.9747342261493857, | |
| "grad_norm": 13.469632148742676, | |
| "learning_rate": 3.3835116781728565e-10, | |
| "loss": 0.8785, | |
| "step": 7060 | |
| }, | |
| { | |
| "epoch": 0.9761148695292006, | |
| "grad_norm": 14.966092109680176, | |
| "learning_rate": 3.0259130519630915e-10, | |
| "loss": 0.8801, | |
| "step": 7070 | |
| }, | |
| { | |
| "epoch": 0.9774955129090156, | |
| "grad_norm": 10.650399208068848, | |
| "learning_rate": 2.6882528225570246e-10, | |
| "loss": 0.8797, | |
| "step": 7080 | |
| }, | |
| { | |
| "epoch": 0.9788761562888306, | |
| "grad_norm": 11.630505561828613, | |
| "learning_rate": 2.370537742791945e-10, | |
| "loss": 0.8898, | |
| "step": 7090 | |
| }, | |
| { | |
| "epoch": 0.9802567996686455, | |
| "grad_norm": 8.717507362365723, | |
| "learning_rate": 2.0727741666235476e-10, | |
| "loss": 0.9037, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 0.9816374430484606, | |
| "grad_norm": 13.472414016723633, | |
| "learning_rate": 1.794968048999368e-10, | |
| "loss": 0.8784, | |
| "step": 7110 | |
| }, | |
| { | |
| "epoch": 0.9830180864282756, | |
| "grad_norm": 16.761281967163086, | |
| "learning_rate": 1.537124945739321e-10, | |
| "loss": 0.8637, | |
| "step": 7120 | |
| }, | |
| { | |
| "epoch": 0.9843987298080906, | |
| "grad_norm": 11.78791618347168, | |
| "learning_rate": 1.299250013425124e-10, | |
| "loss": 0.897, | |
| "step": 7130 | |
| }, | |
| { | |
| "epoch": 0.9857793731879055, | |
| "grad_norm": 12.46883773803711, | |
| "learning_rate": 1.0813480092963789e-10, | |
| "loss": 0.8981, | |
| "step": 7140 | |
| }, | |
| { | |
| "epoch": 0.9871600165677206, | |
| "grad_norm": 9.010354995727539, | |
| "learning_rate": 8.834232911560934e-11, | |
| "loss": 0.8751, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 0.9885406599475356, | |
| "grad_norm": 17.35738182067871, | |
| "learning_rate": 7.054798172833053e-11, | |
| "loss": 0.8696, | |
| "step": 7160 | |
| }, | |
| { | |
| "epoch": 0.9899213033273505, | |
| "grad_norm": 14.424629211425781, | |
| "learning_rate": 5.4752114635370215e-11, | |
| "loss": 0.8592, | |
| "step": 7170 | |
| }, | |
| { | |
| "epoch": 0.9913019467071655, | |
| "grad_norm": 9.579642295837402, | |
| "learning_rate": 4.095504373688996e-11, | |
| "loss": 0.8883, | |
| "step": 7180 | |
| }, | |
| { | |
| "epoch": 0.9926825900869806, | |
| "grad_norm": 13.51734733581543, | |
| "learning_rate": 2.915704495929372e-11, | |
| "loss": 0.857, | |
| "step": 7190 | |
| }, | |
| { | |
| "epoch": 0.9940632334667955, | |
| "grad_norm": 35.478233337402344, | |
| "learning_rate": 1.9358354249721098e-11, | |
| "loss": 0.9121, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 0.9954438768466105, | |
| "grad_norm": 10.26551628112793, | |
| "learning_rate": 1.1559167571317807e-11, | |
| "loss": 0.8941, | |
| "step": 7210 | |
| }, | |
| { | |
| "epoch": 0.9968245202264255, | |
| "grad_norm": 10.065427780151367, | |
| "learning_rate": 5.759640899327678e-12, | |
| "loss": 0.8778, | |
| "step": 7220 | |
| }, | |
| { | |
| "epoch": 0.9982051636062405, | |
| "grad_norm": 12.31423282623291, | |
| "learning_rate": 1.959890217984039e-12, | |
| "loss": 0.885, | |
| "step": 7230 | |
| }, | |
| { | |
| "epoch": 0.9995858069860555, | |
| "grad_norm": 14.609333992004395, | |
| "learning_rate": 1.5999151815604406e-13, | |
| "loss": 0.8856, | |
| "step": 7240 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 7243, | |
| "total_flos": 8.406253259599538e+19, | |
| "train_loss": 0.9649970227309589, | |
| "train_runtime": 70078.4596, | |
| "train_samples_per_second": 6.614, | |
| "train_steps_per_second": 0.103 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 7243, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 8.406253259599538e+19, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |