Instructions to use modrill/MN9-SHORT-515K with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use modrill/MN9-SHORT-515K with PEFT:
from peft import PeftModel from transformers import AutoModelForCausalLM base_model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen3-4B-Base") model = PeftModel.from_pretrained(base_model, "modrill/MN9-SHORT-515K") - Notebooks
- Google Colab
- Kaggle
| { | |
| "arm_id": "MN9-A-RR3C-SHORT", | |
| "completed_at_utc": "2026-08-10T06:57:01Z", | |
| "data": { | |
| "length": { | |
| "max": 2043, | |
| "min": 42, | |
| "p50": 1249, | |
| "p95": 1929 | |
| }, | |
| "max_mn8_loss_active_tokens": 2044, | |
| "realized_ratios": { | |
| "S0": 0.099736, | |
| "S1": 0.749915, | |
| "S2": 0.150349 | |
| }, | |
| "rows": 441, | |
| "target_ratios": { | |
| "S0": 0.1, | |
| "S1": 0.75, | |
| "S2": 0.15 | |
| } | |
| }, | |
| "evaluation": { | |
| "accuracy": 0.15, | |
| "by_seed": { | |
| "42": { | |
| "correct": 9, | |
| "n": 60 | |
| }, | |
| "43": { | |
| "correct": 8, | |
| "n": 60 | |
| }, | |
| "44": { | |
| "correct": 8, | |
| "n": 60 | |
| }, | |
| "45": { | |
| "correct": 11, | |
| "n": 60 | |
| } | |
| }, | |
| "correct": 36, | |
| "dynamic_lora": true, | |
| "metrics": { | |
| "accuracy": 0.15, | |
| "by_seed": { | |
| "42": { | |
| "accuracy": 0.15, | |
| "correct": 9, | |
| "n": 60 | |
| }, | |
| "43": { | |
| "accuracy": 0.13333333333333333, | |
| "correct": 8, | |
| "n": 60 | |
| }, | |
| "44": { | |
| "accuracy": 0.13333333333333333, | |
| "correct": 8, | |
| "n": 60 | |
| }, | |
| "45": { | |
| "accuracy": 0.18333333333333332, | |
| "correct": 11, | |
| "n": 60 | |
| } | |
| }, | |
| "correct": 36, | |
| "n": 240, | |
| "response_tokens": { | |
| "mean": 4180.758333333333, | |
| "p50": 2013.5, | |
| "p95": 8192 | |
| }, | |
| "tokens_per_correct": 27871.722222222223 | |
| }, | |
| "n": 240, | |
| "score_authority": "EvalScope reviews", | |
| "seeds": [ | |
| 42, | |
| 43, | |
| 44, | |
| 45 | |
| ], | |
| "suites": [ | |
| "aime24", | |
| "aime25" | |
| ], | |
| "vs_BaseFalse_21_240": { | |
| "basefalse_correct": 21, | |
| "delta_correct": 15 | |
| }, | |
| "vs_RR3C_36_240_reference_only": { | |
| "beat_gate": false, | |
| "delta_correct": 0, | |
| "rr3c_correct": 36 | |
| } | |
| }, | |
| "key_paths": { | |
| "arm_dir": "/workspace/DDC-v3.1/runs/mn9_rr3_short_4arm_20260810/arms/MN9-A-RR3C-SHORT", | |
| "eval_root": "/workspace/DDC-v3.1/runs/mn9_rr3_short_4arm_20260810/eval/MN9-A-RR3C-SHORT", | |
| "milestone_128k": "/workspace/DDC-v3.1/runs/mn9_rr3_short_4arm_20260810/arms/MN9-A-RR3C-SHORT/milestones/128K", | |
| "milestone_515k": "/workspace/DDC-v3.1/runs/mn9_rr3_short_4arm_20260810/arms/MN9-A-RR3C-SHORT/milestones/515K", | |
| "training_config": "/workspace/DDC-v3.1/runs/mn9_rr3_short_4arm_20260810/arms/MN9-A-RR3C-SHORT/TRAINING_CONFIG.json", | |
| "update_plan": "/workspace/DDC-v3.1/runs/mn9_rr3_short_4arm_20260810/arms/MN9-A-RR3C-SHORT/UPDATE_PLAN.json" | |
| }, | |
| "mode": "nothink", | |
| "schema": "ddc-mn9-rr3-short-4arm-result-card/1", | |
| "status": "DONE", | |
| "training": { | |
| "adapter_path": "/workspace/DDC-v3.1/runs/mn9_rr3_short_4arm_20260810/arms/MN9-A-RR3C-SHORT/milestones/515K/adapter", | |
| "adapter_sha256": "683a178fd516323c92ade9b1c3e4f8bc3551f774ceabafbc83e7f138052f49b1", | |
| "completed_optimizer_updates": 16, | |
| "learning_rate": 2e-05, | |
| "no_merge_or_bake": true, | |
| "pure_base": true, | |
| "realized_active_tokens": 534970, | |
| "scheduler": "constant", | |
| "warmup_steps": 0 | |
| } | |
| } | |