84 lines
2.1 KiB
JSON
84 lines
2.1 KiB
JSON
|
{
|
||
|
|
"total_train": 6396,
|
||
|
|
"raw_before_dedup": 6396,
|
||
|
|
"config": {
|
||
|
|
"model": {
|
||
|
|
"id": "Qwen/Qwen3-1.7B",
|
||
|
|
"cache_dir": "/scratch/hf_cache"
|
||
|
|
},
|
||
|
|
"system_prompt": "You are a knowledgeable assistant. For multiple-choice questions, reason step by step, then write only the letter of the correct answer inside \\boxed{}.\n",
|
||
|
|
"data": {
|
||
|
|
"sources": [
|
||
|
|
"/scratch/data/gk_jennifer_knowledge_v6.jsonl"
|
||
|
|
],
|
||
|
|
"max_samples": null,
|
||
|
|
"sampling_strategy": "source_balanced",
|
||
|
|
"max_seq_length": 4096,
|
||
|
|
"max_think_chars": 0,
|
||
|
|
"seed": 7
|
||
|
|
},
|
||
|
|
"training": {
|
||
|
|
"output_dir": "/scratch/checkpoints/gk_thinking_v6_knowledge_base",
|
||
|
|
"num_train_epochs": 2,
|
||
|
|
"learning_rate": 7e-06,
|
||
|
|
"per_device_train_batch_size": 2,
|
||
|
|
"gradient_accumulation_steps": 8,
|
||
|
|
"warmup_ratio": 0.05,
|
||
|
|
"seed": 7,
|
||
|
|
"bf16": true,
|
||
|
|
"logging_steps": 10,
|
||
|
|
"save_strategy": "epoch",
|
||
|
|
"gradient_checkpointing": true
|
||
|
|
},
|
||
|
|
"generation": {
|
||
|
|
"do_sample": true,
|
||
|
|
"temperature": 0.6,
|
||
|
|
"top_k": 20,
|
||
|
|
"top_p": 0.95,
|
||
|
|
"eos_token_id": [
|
||
|
|
151645,
|
||
|
|
151643
|
||
|
|
],
|
||
|
|
"pad_token_id": 151643
|
||
|
|
}
|
||
|
|
},
|
||
|
|
"source_counts": {
|
||
|
|
"cot_distill/jennifer/other": 201,
|
||
|
|
"cot_distill/jennifer/business": 91,
|
||
|
|
"cot_distill/jennifer/mmlu": 4602,
|
||
|
|
"cot_distill/jennifer/biology": 140,
|
||
|
|
"cot_distill/jennifer/engineering": 49,
|
||
|
|
"cot_distill/jennifer/economics": 203,
|
||
|
|
"cot_distill/jennifer/math": 196,
|
||
|
|
"cot_distill/jennifer/health": 170,
|
||
|
|
"cot_distill/jennifer/physics": 118,
|
||
|
|
"cot_distill/jennifer/law": 150,
|
||
|
|
"cot_distill/jennifer/psychology": 161,
|
||
|
|
"cot_distill/jennifer/computer science": 60,
|
||
|
|
"cot_distill/jennifer/history": 56,
|
||
|
|
"cot_distill/jennifer/philosophy": 93,
|
||
|
|
"cot_distill/jennifer/chemistry": 106
|
||
|
|
},
|
||
|
|
"answer_letter_counts": {
|
||
|
|
"A": 1364,
|
||
|
|
"B": 1414,
|
||
|
|
"C": 1456,
|
||
|
|
"D": 1185,
|
||
|
|
"E": 383,
|
||
|
|
"F": 115,
|
||
|
|
"G": 101,
|
||
|
|
"H": 80,
|
||
|
|
"I": 76,
|
||
|
|
"J": 50,
|
||
|
|
"K": 29,
|
||
|
|
"L": 36,
|
||
|
|
"M": 32,
|
||
|
|
"N": 19,
|
||
|
|
"O": 16,
|
||
|
|
"P": 9,
|
||
|
|
"Q": 9,
|
||
|
|
"R": 13,
|
||
|
|
"S": 5,
|
||
|
|
"T": 4
|
||
|
|
}
|
||
|
|
}
|