Instructions to use hunterbown/trainless-math-2b-loop-repair with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use hunterbown/trainless-math-2b-loop-repair with PEFT:
from peft import PeftModel from transformers import AutoModelForCausalLM base_model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen3.5-2B") model = PeftModel.from_pretrained(base_model, "hunterbown/trainless-math-2b-loop-repair") - Notebooks
- Google Colab
- Kaggle
File size: 5,492 Bytes
fa89a16 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 | {
"schema": "trainless-math-evaluation/1",
"scope": "Complete existing 200-item verification slice; requalification of exact analytic PEFT, not a newly held-out discovery",
"model": {
"base_model": "Qwen/Qwen3.5-2B",
"base_revision": "15852e8c16360a2fea060d615a32b45270f8a8fc",
"adapter_model_sha256": "f67ffbbc22c7a0c0e1dd9ce28459751b18ab91de6132390a164c611f5744e27b",
"adapter_config_sha256": "1cf3e3e143f8447c72287aa37419f1661b9afae6e8f7ba1bb848bed95a531752",
"dtype": "bfloat16"
},
"dataset": {
"id": "HuggingFaceH4/MATH-500",
"offset_zero_based": 64,
"last_index_inclusive": 263,
"n": 200,
"selected_item_sha256": "8a3b16a62f8b0755ebbd49d4fa3b3cd05a3dba3fa17e234f6001e92b3c36f6a1",
"selected_hash_algorithm": "sha256(json.dumps(items, sort_keys=True).encode()); default spaces and ensure_ascii=True; each item has only id, problem, answer",
"historical_hub_revision": null,
"historical_revision_recorded": false,
"current_upstream_verification": {
"verified_at": "2026-11-10T07:36:38Z",
"current_upstream_revision": "6e4ed1a2a79af7d8630a6b768ec859cb5af4d3be",
"license_source_revision": "7ecc794703b2877f63226f2477a49b34f9b25163",
"upstream_test_jsonl_sha256": "35dc41080a3680858b27fa7e0533d2d547825316fc5dafe5d316f4ccc5a06132",
"current_upstream_selected_sha256": "8a3b16a62f8b0755ebbd49d4fa3b3cd05a3dba3fa17e234f6001e92b3c36f6a1"
}
},
"generation": {
"output_token_cap": 32768,
"max_num_seqs": 64,
"sampling": {
"temperature": 0.6,
"top_p": 0.95,
"top_k": 20,
"presence_penalty": 1.5,
"seed": 0,
"thinking": true
},
"runtime": {
"vllm": "0.31.0",
"torch": "2.13.0",
"transformers": "5.17.0",
"safetensors": "0.8.0",
"numpy": "2.2.6"
},
"gpu": "NVIDIA H100 PCIe"
},
"prompt_reconstruction": {
"method": "Reconstructed with the same pinned tokenizer and recorded historical Transformers version; original generation JSON did not store prompt strings",
"historical_prompts_stored": false,
"base_model": "Qwen/Qwen3.5-2B",
"base_revision": "15852e8c16360a2fea060d615a32b45270f8a8fc",
"transformers": "5.17.0",
"tokenizer_class": "Qwen2Tokenizer",
"tokenizer_config_sha256": "49e2b6e395f959f077f1e992b338919c0d4a9732fc6e613995e06557f843500c",
"chat_template_sha256": "273d8e0e683b885071fb17e08d71e5f2a5ddfb5309756181681de4f5a1822d80",
"prompt_suffix": "Please reason step by step, and put your final answer within \\boxed{}.",
"thinking": true
},
"grading": {
"description": "repository evals.graders.for_benchmark(math500), unchanged",
"runtime": {
"python": "3.10.12",
"math-verify": "0.9.0",
"sympy": "1.12"
},
"source_sha256": {
"graders.py": "1cfec08ffbfaf9b95e25c87c2c4ea94847f0e47a149e279598df09d88ce1908c",
"benchmarks.py": "d69ac797dd546c6ab29dadfc516936199d1008f1271ea3514b6465db30619830",
"loops.py": "c93ab48e93ccfa18689d4b13a7dff3fb7a55389be528fb8576cb27b806e34e51"
}
},
"original_source_sha256": {
"generations.json": "868f8118474c7acf2361bfbbed29dc8c50d9813c395e269f4bad4e59854b02ff",
"graded.json": "d255d4ea60870bf9450cecff8ce909fce2b9b4a22fc94fbba76d03d48b7cf5c2",
"qualification-metadata.json": "3d334352fc4dba360b79174646a6f2d9fb259bfc79667f4909f768d28e9aacf4",
"reference_items_file": "c5198b451c6f1ca6bf4fcfbbac86161bd9a52442c6983fac60850a636b9eb680",
"prompt_reconstruction_file": "18162f85fd53db147cd810b84a91eb7ed4b35859717c1e35ca006368c1c9e2d8",
"attribution_file": "2a43117ca8354f96654ca990295ab1c490b6845a20a413223d3050dd1cb9bef4",
"original_benchmark_cache": "0baf4950c69ab971f1b45437b965927e5a2a8fda0d2c884c408c9b9c087fca84",
"generation_driver": "4cbe1f202744ece0c57aa2b7e4d5e3aa1235fa27e2096595aedccc5996a221ff"
},
"results": {
"base": {
"n": 200,
"correct": 136,
"accuracy": 0.68,
"wilson95": [
0.6124810156588646,
0.7407346668453195
],
"capped": 63,
"total_output_tokens": 3263048,
"generation_wall_seconds": 759.5006894249236
},
"adapter": {
"n": 200,
"correct": 152,
"accuracy": 0.76,
"wilson95": [
0.6962665902079583,
0.8139338400758631
],
"capped": 48,
"total_output_tokens": 2863410,
"generation_wall_seconds": 760.9235217790119
}
},
"paired": {
"accuracy_delta": 0.08,
"bootstrap95": [
0.03,
0.135
],
"bootstrap_repetitions": 20000,
"seed": 20261110,
"base_wrong_adapter_right": 23,
"base_right_adapter_wrong": 7,
"note": "paired item bootstrap, same saved questions; model sampling variance from additional seeds unmeasured"
},
"limitations": [
"Same previously evaluated questions; no new held-out discovery claim.",
"One sampling seed; seed variance unmeasured.",
"No wall-clock speedup established; both generation arms took about 760 seconds.",
"Only the exact analytic adapter/runtime is measured; merged and GGUF variants do not inherit the score.",
"Historical Hub dataset revision was not saved. Current upstream identity was verified separately.",
"Prompt strings were reconstructed from the pinned tokenizer and original runtime, not recovered from the generation JSON.",
"Automatic answer labels can miss contradictions; full outputs are included for human review."
]
}
|