47 lines
4.7 KiB
JSON
47 lines
4.7 KiB
JSON
|
{
|
||
|
|
"schema_version": 1,
|
||
|
|
"status": "frozen_before_training_and_validation_ranking",
|
||
|
|
"frozen_at_kst": "2026-07-15T09:01:45+09:00",
|
||
|
|
"seed": 42,
|
||
|
|
"common": {
|
||
|
|
"optimizer_steps": 900,
|
||
|
|
"effective_batch_size": 16,
|
||
|
|
"per_device_train_batch_size": 1,
|
||
|
|
"gradient_accumulation_steps": 16,
|
||
|
|
"gradient_checkpointing": true,
|
||
|
|
"bf16": true,
|
||
|
|
"attn_implementation": "sdpa",
|
||
|
|
"cudnn_sdpa": false,
|
||
|
|
"lr_scheduler_type": "cosine",
|
||
|
|
"max_length": 2048,
|
||
|
|
"max_prompt_length": 1800,
|
||
|
|
"save_total_limit": 1,
|
||
|
|
"wandb_entity": "promotion-kim",
|
||
|
|
"wandb_project": "mnpo",
|
||
|
|
"wandb_group": "qwen3-8b-fair-demo-20260715"
|
||
|
|
},
|
||
|
|
"budget_rule": "Exactly two configs per reported method; no extension after validation rankings are visible.",
|
||
|
|
"candidates": [
|
||
|
|
{"id":"ronpo_full_a","method":"ronpo_full_expect","dataset":"ronpo_full_expect_kall","learning_rate":1e-7,"warmup_ratio":0.15,"ronpo_alpha":0.25,"ronpo_tau":0.10,"reference_anchor_weight":0.075,"preference_sft_weight":0.0075,"expected_support_k":0},
|
||
|
|
{"id":"ronpo_full_b","method":"ronpo_full_expect","dataset":"ronpo_full_expect_k6","learning_rate":2.5e-7,"warmup_ratio":0.20,"ronpo_alpha":0.35,"ronpo_tau":0.05,"reference_anchor_weight":0.10,"preference_sft_weight":0.01,"expected_support_k":6},
|
||
|
|
{"id":"ronpo_top_a","method":"ronpo_k_only","dataset":"ronpo_k_only_k1","learning_rate":5e-8,"warmup_ratio":0.20,"ronpo_alpha":0.15,"ronpo_tau":0.10,"reference_anchor_weight":0.10,"preference_sft_weight":0.01,"expected_support_k":1},
|
||
|
|
{"id":"ronpo_top_b","method":"ronpo_k_only","dataset":"ronpo_k_only_k2","learning_rate":1e-7,"warmup_ratio":0.15,"ronpo_alpha":0.25,"ronpo_tau":0.10,"reference_anchor_weight":0.075,"preference_sft_weight":0.0075,"expected_support_k":2},
|
||
|
|
{"id":"dpo_a","method":"dpo","dataset":"avg","learning_rate":2.5e-7,"warmup_ratio":0.10,"dpo_beta":0.05,"reference_anchor_weight":0.02,"preference_sft_weight":0.002},
|
||
|
|
{"id":"dpo_b","method":"dpo","dataset":"avg","learning_rate":5e-7,"warmup_ratio":0.15,"dpo_beta":0.10,"reference_anchor_weight":0.05,"preference_sft_weight":0.005},
|
||
|
|
{"id":"ipo_a","method":"ipo","dataset":"avg","learning_rate":2.5e-7,"warmup_ratio":0.15,"dpo_beta":0.20,"reference_anchor_weight":0.05,"preference_sft_weight":0.005},
|
||
|
|
{"id":"ipo_b","method":"ipo","dataset":"avg","learning_rate":5e-7,"warmup_ratio":0.20,"dpo_beta":0.50,"reference_anchor_weight":0.075,"preference_sft_weight":0.0075},
|
||
|
|
{"id":"simpo_a","method":"simpo","dataset":"avg","learning_rate":2.5e-7,"warmup_ratio":0.10,"simpo_beta":1.0,"simpo_gamma":0.3,"reference_anchor_weight":0.02,"preference_sft_weight":0.002},
|
||
|
|
{"id":"simpo_b","method":"simpo","dataset":"avg","learning_rate":5e-7,"warmup_ratio":0.15,"simpo_beta":2.0,"simpo_gamma":0.6,"reference_anchor_weight":0.05,"preference_sft_weight":0.005},
|
||
|
|
{"id":"sppo_a","method":"sppo_avg","dataset":"avg","learning_rate":2.5e-7,"warmup_ratio":0.15,"eta":0.005,"reference_anchor_weight":0.05,"preference_sft_weight":0.005},
|
||
|
|
{"id":"sppo_b","method":"sppo_avg","dataset":"avg","learning_rate":5e-7,"warmup_ratio":0.20,"eta":0.010,"reference_anchor_weight":0.075,"preference_sft_weight":0.0075},
|
||
|
|
{"id":"inpo_a","method":"inpo_avg","dataset":"avg","learning_rate":2.5e-7,"warmup_ratio":0.15,"eta":0.050,"ratio":0.3333,"reference_anchor_weight":0.05,"preference_sft_weight":0.005},
|
||
|
|
{"id":"inpo_b","method":"inpo_avg","dataset":"avg","learning_rate":5e-7,"warmup_ratio":0.20,"eta":0.100,"ratio":0.3333,"reference_anchor_weight":0.075,"preference_sft_weight":0.0075},
|
||
|
|
{"id":"ht_help_a","method":"ht_mnpo_helpfulness","dataset":"ht_mnpo_helpfulness","learning_rate":2.5e-7,"warmup_ratio":0.10,"eta":0.005,"reference_anchor_weight":0.02,"preference_sft_weight":0.002},
|
||
|
|
{"id":"ht_help_b","method":"ht_mnpo_helpfulness","dataset":"ht_mnpo_helpfulness","learning_rate":5e-7,"warmup_ratio":0.15,"eta":0.010,"reference_anchor_weight":0.05,"preference_sft_weight":0.005},
|
||
|
|
{"id":"ht_safety_a","method":"ht_mnpo_safety","dataset":"ht_mnpo_safety","learning_rate":2.5e-7,"warmup_ratio":0.10,"eta":0.005,"reference_anchor_weight":0.02,"preference_sft_weight":0.002},
|
||
|
|
{"id":"ht_safety_b","method":"ht_mnpo_safety","dataset":"ht_mnpo_safety","learning_rate":5e-7,"warmup_ratio":0.15,"eta":0.010,"reference_anchor_weight":0.05,"preference_sft_weight":0.005},
|
||
|
|
{"id":"ht_concise_a","method":"ht_mnpo_conciseness","dataset":"ht_mnpo_conciseness","learning_rate":2.5e-7,"warmup_ratio":0.10,"eta":0.005,"reference_anchor_weight":0.02,"preference_sft_weight":0.002},
|
||
|
|
{"id":"ht_concise_b","method":"ht_mnpo_conciseness","dataset":"ht_mnpo_conciseness","learning_rate":5e-7,"warmup_ratio":0.15,"eta":0.010,"reference_anchor_weight":0.05,"preference_sft_weight":0.005}
|
||
|
|
]
|
||
|
|
}
|