Text Classification
Transformers
Safetensors
English
qwen3_5_text
text-generation
decision-model
system-one
typed-decisions
calibrated
decider
Instructions to use mlboydaisuke/decider-2b-coreai-ft with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use mlboydaisuke/decider-2b-coreai-ft with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-classification", model="mlboydaisuke/decider-2b-coreai-ft")# pip install -U transformers accelerate # Load model directly from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("mlboydaisuke/decider-2b-coreai-ft") model = AutoModelForCausalLM.from_pretrained("mlboydaisuke/decider-2b-coreai-ft", device_map="auto") - Notebooks
- Google Colab
- Kaggle
v3: decider-2b v11 + 6,362 verified typed-decision items incl. 543 long documents (delta recipe, 3 epochs, lr 8e-6); card numbers from the 2026-09-26 run
284dfd3 verified Download eval/train.log from mlboydaisuke/decider-2b-coreai-ft: direct link, hf CLI and curl.
- Browser
- Download file 8.33 kB
-
https://huggingface.co/mlboydaisuke/decider-2b-coreai-ft/resolve/main/eval/train.log
- Command line
-
hf download hf://mlboydaisuke/decider-2b-coreai-ft/eval/train.log
-
curl -L -o train.log https://huggingface.co/mlboydaisuke/decider-2b-coreai-ft/resolve/main/eval/train.log
8.33 kB
| [args] {"model": "Mapika/decider-2b", "data": "/content/data/coreai.pkl", "out": "/content/runs/delta", "epochs": 3.0, "lr": 8e-06, "wd": 0.0, "warmup": 20, "max_tokens": 12288, "accum": 2, "max_ctx": 16384, "brier_w": 0.0, "label_smooth": 0.0, "eval_every": 100000, "eval_limit": 100, "train_cap": 0, "seed": 0, "resample_every_epoch": true, "none_prob": 0.0, "schema_first_prob": 0.0, "max_options": 255} | |
| [data] train examples 6362; tokenizing... | |
| [data] 6362 items, 6362 questions, 10.9M tokens, tokenized in 28s | |
| [sched] 515 optimizer steps/epoch, 1545 total | |
| [train] step 20/1545 ep 0 ce 0.8879 gn 74.00 lr 8.00e-06 3.7min eta 284min 1940tok/s mem 20GB | |
| [train] step 40/1545 ep 0 ce 0.7011 gn 136.00 lr 8.00e-06 4.6min eta 69min 8256tok/s mem 20GB | |
| [train] step 60/1545 ep 0 ce 0.6629 gn 100.00 lr 7.99e-06 5.5min eta 68min 8023tok/s mem 20GB | |
| [train] step 80/1545 ep 0 ce 0.5311 gn 127.00 lr 7.97e-06 6.4min eta 63min 8794tok/s mem 20GB | |
| [train] step 100/1545 ep 0 ce 0.7026 gn 113.00 lr 7.95e-06 7.3min eta 62min 8640tok/s mem 20GB | |
| [train] step 120/1545 ep 0 ce 0.5571 gn 69.00 lr 7.92e-06 8.1min eta 61min 8762tok/s mem 20GB | |
| [train] step 140/1545 ep 0 ce 0.7724 gn 84.00 lr 7.88e-06 9.0min eta 61min 8448tok/s mem 20GB | |
| [train] step 160/1545 ep 0 ce 0.6581 gn 154.00 lr 7.83e-06 9.8min eta 59min 8911tok/s mem 20GB | |
| [train] step 180/1545 ep 0 ce 0.5204 gn 45.25 lr 7.78e-06 10.7min eta 61min 8208tok/s mem 20GB | |
| [train] step 200/1545 ep 0 ce 0.5297 gn 54.00 lr 7.73e-06 11.6min eta 59min 8508tok/s mem 20GB | |
| [train] step 220/1545 ep 0 ce 0.5509 gn 42.00 lr 7.67e-06 12.5min eta 55min 8601tok/s mem 20GB | |
| [train] step 240/1545 ep 0 ce 0.7435 gn 50.50 lr 7.60e-06 13.3min eta 56min 8717tok/s mem 20GB | |
| [train] step 260/1545 ep 0 ce 0.5157 gn 107.00 lr 7.52e-06 14.2min eta 55min 8709tok/s mem 20GB | |
| [train] step 280/1545 ep 0 ce 0.6843 gn 147.00 lr 7.44e-06 15.0min eta 54min 8624tok/s mem 20GB | |
| [train] step 300/1545 ep 0 ce 0.5498 gn 76.50 lr 7.35e-06 15.9min eta 53min 8609tok/s mem 20GB | |
| [train] step 320/1545 ep 0 ce 0.6019 gn 294.00 lr 7.26e-06 16.7min eta 51min 8647tok/s mem 20GB | |
| [train] step 340/1545 ep 0 ce 0.5030 gn 81.00 lr 7.16e-06 17.5min eta 50min 8658tok/s mem 20GB | |
| [train] step 360/1545 ep 0 ce 0.5059 gn 68.50 lr 7.06e-06 18.4min eta 50min 8506tok/s mem 20GB | |
| [train] step 380/1545 ep 0 ce 0.4691 gn 59.75 lr 6.95e-06 19.2min eta 48min 8813tok/s mem 20GB | |
| [train] step 400/1545 ep 0 ce 0.5794 gn 37.00 lr 6.84e-06 20.0min eta 48min 8746tok/s mem 20GB | |
| [train] step 420/1545 ep 0 ce 0.4682 gn 135.00 lr 6.72e-06 20.9min eta 48min 8628tok/s mem 20GB | |
| [train] step 440/1545 ep 0 ce 0.5367 gn 27.75 lr 6.59e-06 21.7min eta 47min 8686tok/s mem 20GB | |
| [train] step 460/1545 ep 0 ce 0.5448 gn 122.00 lr 6.47e-06 22.6min eta 45min 8779tok/s mem 20GB | |
| [train] step 480/1545 ep 0 ce 0.5098 gn 98.00 lr 6.33e-06 23.4min eta 45min 8563tok/s mem 20GB | |
| [train] step 500/1545 ep 0 ce 0.3887 gn 110.50 lr 6.20e-06 24.3min eta 44min 8392tok/s mem 20GB | |
| [train] step 520/1545 ep 1 ce 0.4085 gn 22.50 lr 6.06e-06 25.8min eta 77min 4948tok/s mem 20GB | |
| [train] step 540/1545 ep 1 ce 0.2176 gn 85.50 lr 5.92e-06 26.6min eta 43min 8637tok/s mem 20GB | |
| [train] step 560/1545 ep 1 ce 0.3074 gn 99.00 lr 5.77e-06 27.5min eta 42min 8626tok/s mem 20GB | |
| [train] step 580/1545 ep 1 ce 0.1576 gn 35.00 lr 5.62e-06 28.3min eta 42min 8643tok/s mem 20GB | |
| [train] step 600/1545 ep 1 ce 0.2292 gn 73.50 lr 5.47e-06 29.2min eta 40min 8502tok/s mem 20GB | |
| [train] step 620/1545 ep 1 ce 0.1425 gn 73.00 lr 5.31e-06 30.0min eta 40min 8695tok/s mem 20GB | |
| [train] step 640/1545 ep 1 ce 0.1507 gn 22.38 lr 5.16e-06 30.9min eta 39min 8543tok/s mem 20GB | |
| [train] step 660/1545 ep 1 ce 0.2277 gn 44.25 lr 5.00e-06 31.7min eta 37min 8738tok/s mem 20GB | |
| [train] step 680/1545 ep 1 ce 0.3366 gn 3.41 lr 4.84e-06 32.5min eta 36min 8654tok/s mem 20GB | |
| [train] step 700/1545 ep 1 ce 0.2382 gn 59.25 lr 4.68e-06 33.4min eta 36min 8660tok/s mem 20GB | |
| [train] step 720/1545 ep 1 ce 0.2296 gn 11.00 lr 4.51e-06 34.3min eta 36min 8664tok/s mem 20GB | |
| [train] step 740/1545 ep 1 ce 0.1894 gn 93.50 lr 4.35e-06 35.1min eta 35min 8703tok/s mem 20GB | |
| [train] step 760/1545 ep 1 ce 0.2151 gn 310.00 lr 4.19e-06 36.0min eta 34min 8598tok/s mem 20GB | |
| [train] step 780/1545 ep 1 ce 0.0980 gn 0.61 lr 4.02e-06 36.8min eta 32min 8575tok/s mem 20GB | |
| [train] step 800/1545 ep 1 ce 0.2232 gn 70.00 lr 3.86e-06 37.6min eta 31min 8746tok/s mem 20GB | |
| [train] step 820/1545 ep 1 ce 0.2448 gn 32.75 lr 3.69e-06 38.5min eta 30min 8608tok/s mem 20GB | |
| [train] step 840/1545 ep 1 ce 0.1860 gn 3.02 lr 3.53e-06 39.3min eta 30min 8670tok/s mem 20GB | |
| [train] step 860/1545 ep 1 ce 0.1953 gn 208.00 lr 3.36e-06 40.2min eta 29min 8810tok/s mem 20GB | |
| [train] step 880/1545 ep 1 ce 0.3090 gn 5.84 lr 3.20e-06 41.0min eta 28min 8754tok/s mem 20GB | |
| [train] step 900/1545 ep 1 ce 0.1677 gn 148.00 lr 3.04e-06 41.8min eta 27min 8746tok/s mem 20GB | |
| [train] step 920/1545 ep 1 ce 0.1698 gn 32.00 lr 2.88e-06 42.7min eta 27min 8412tok/s mem 20GB | |
| [train] step 940/1545 ep 1 ce 0.2119 gn 225.00 lr 2.72e-06 43.6min eta 26min 8684tok/s mem 20GB | |
| [train] step 960/1545 ep 1 ce 0.1439 gn 193.00 lr 2.57e-06 44.4min eta 25min 8686tok/s mem 20GB | |
| [train] step 980/1545 ep 1 ce 0.1796 gn 18.50 lr 2.42e-06 45.3min eta 24min 8560tok/s mem 20GB | |
| [train] step 1000/1545 ep 1 ce 0.2553 gn 68.50 lr 2.27e-06 46.1min eta 23min 8685tok/s mem 20GB | |
| [train] step 1020/1545 ep 1 ce 0.2300 gn 110.50 lr 2.12e-06 47.0min eta 22min 8932tok/s mem 20GB | |
| [train] step 1040/1545 ep 2 ce 0.0862 gn 33.00 lr 1.98e-06 48.5min eta 39min 4722tok/s mem 20GB | |
| [train] step 1060/1545 ep 2 ce 0.1130 gn 106.50 lr 1.84e-06 49.3min eta 20min 8767tok/s mem 20GB | |
| [train] step 1080/1545 ep 2 ce 0.0386 gn 0.01 lr 1.70e-06 50.2min eta 20min 8638tok/s mem 20GB | |
| [train] step 1100/1545 ep 2 ce 0.0717 gn 33.75 lr 1.57e-06 51.0min eta 19min 8790tok/s mem 20GB | |
| [train] step 1120/1545 ep 2 ce 0.0447 gn 13.38 lr 1.44e-06 51.9min eta 18min 8591tok/s mem 20GB | |
| [train] step 1140/1545 ep 2 ce 0.1981 gn 0.03 lr 1.31e-06 52.7min eta 17min 8626tok/s mem 20GB | |
| [train] step 1160/1545 ep 2 ce 0.1301 gn 107.50 lr 1.19e-06 53.6min eta 16min 8833tok/s mem 20GB | |
| [train] step 1180/1545 ep 2 ce 0.0541 gn 0.02 lr 1.08e-06 54.4min eta 15min 8582tok/s mem 20GB | |
| [train] step 1200/1545 ep 2 ce 0.0595 gn 0.04 lr 9.68e-07 55.3min eta 15min 8781tok/s mem 20GB | |
| [train] step 1220/1545 ep 2 ce 0.0695 gn 54.75 lr 8.64e-07 56.1min eta 14min 8783tok/s mem 20GB | |
| [train] step 1240/1545 ep 2 ce 0.0858 gn 27.00 lr 7.64e-07 57.0min eta 13min 8472tok/s mem 20GB | |
| [train] step 1260/1545 ep 2 ce 0.0751 gn 0.00 lr 6.70e-07 57.8min eta 12min 8529tok/s mem 20GB | |
| [train] step 1280/1545 ep 2 ce 0.0745 gn 8.31 lr 5.81e-07 58.7min eta 11min 8842tok/s mem 20GB | |
| [train] step 1300/1545 ep 2 ce 0.1348 gn 0.00 lr 4.99e-07 59.5min eta 10min 8645tok/s mem 20GB | |
| [train] step 1320/1545 ep 2 ce 0.0701 gn 12.94 lr 4.22e-07 60.4min eta 9min 8519tok/s mem 20GB | |
| [train] step 1340/1545 ep 2 ce 0.0502 gn 2.12 lr 3.51e-07 61.2min eta 9min 8520tok/s mem 20GB | |
| [train] step 1360/1545 ep 2 ce 0.1445 gn 139.00 lr 2.87e-07 62.1min eta 8min 8803tok/s mem 20GB | |
| [train] step 1380/1545 ep 2 ce 0.1415 gn 0.04 lr 2.29e-07 62.9min eta 7min 8597tok/s mem 20GB | |
| [train] step 1400/1545 ep 2 ce 0.1068 gn 0.04 lr 1.77e-07 63.8min eta 6min 8598tok/s mem 20GB | |
| [train] step 1420/1545 ep 2 ce 0.0716 gn 3.88 lr 1.32e-07 64.6min eta 5min 8702tok/s mem 20GB | |
| [train] step 1440/1545 ep 2 ce 0.0790 gn 0.01 lr 9.32e-08 65.5min eta 4min 8575tok/s mem 20GB | |
| [train] step 1460/1545 ep 2 ce 0.1613 gn 62.25 lr 6.12e-08 66.3min eta 4min 8680tok/s mem 20GB | |
| [train] step 1480/1545 ep 2 ce 0.0795 gn 147.00 lr 3.58e-08 67.2min eta 3min 8674tok/s mem 20GB | |
| [train] step 1500/1545 ep 2 ce 0.1021 gn 0.10 lr 1.72e-08 68.0min eta 2min 8719tok/s mem 20GB | |
| [train] step 1520/1545 ep 2 ce 0.0393 gn 0.01 lr 5.30e-09 68.9min eta 1min 8546tok/s mem 20GB | |
| [train] step 1540/1545 ep 2 ce 0.1235 gn 98.50 lr 2.12e-10 69.7min eta 0min 8705tok/s mem 20GB | |
| [save] /content/runs/delta/model | |
| [eval] coreai_v2_heldout n= 100 acc=0.810 (chance 0.30) nll=0.928 brier=0.325 ece=0.170 aurc=0.066 acc@80=0.863 | |
| [eval-agg] step 1545 {"in_task": {"acc": 0.81, "nll": 0.9282295107841492, "brier": 0.3245173692703247, "ece": 0.1702162915468216, "aurc": 0.06609745285128044, "acc_at_80": 0.8625, "chance": 0.29995238095238097}, "heldout": {}, "n_in": 1, "n_heldout": 0} | |
| [done] saved /content/runs/delta/model | |