code-think-o7b-20260908 / provenance /RUN_IDENTITY.json
modrill's picture
Add provenance/RUN_IDENTITY.json
bdf548e verified
Raw
History Blame Contribute Delete
3.15 kB
{
"schema": "RUN_IDENTITY_V1",
"run_id": "t30b2507-o7b-think-v4-tail151643",
"corrects": "PAIRED_V4",
"model": {
"hf_id": "allenai/Olmo-3-1025-7B",
"revision": "a81bae42db3975be1671e27b9c9a56da1a9f980f",
"local_path": "/workspace/code-sft-infra/models/olmo-3-1025-7b",
"tied_embeddings": false
},
"data": {
"path": "/workspace/code-sft-runs/t30b2507-q4b-tn-v2-lr1e4-e1-20260907/data/THINK_PAIRED_V4_2EP_O7B.parquet",
"rows": 9430,
"sha256": "17554a69f4b99d6d0b0861bece9e959a664084e00dc38ee76a5ba711e2d8da60",
"census": "/workspace/code-sft-runs/t30b2507-q4b-tn-v2-lr1e4-e1-20260907/data/TOKEN_CENSUS_V4_O7B_THINK.json",
"physical_2ep": "concat(selected_v4_table, selected_v4_table) same order; token columns overlaid from TOKEN_CENSUS_V4; O7B also rewrites renderer_id",
"source_1ep_path": "/workspace/code-sft-runs/t30b2507-q4b-tn-v2-lr1e4-e1-20260907/data/THINK_PAIRED_V4.parquet",
"source_1ep_sha256": "cf37c85413f69339508004611e196e42abbc2bddbd3462c209a1b97e1d820002",
"source_1ep_rows": 4715
},
"seeds": {
"train": 42,
"eval": 3407
},
"parameters": {
"epochs": 2,
"target_assistant_tokens": 63378772,
"active_tokens_1ep": 31689386,
"checkpoint_target_tokens": [
31722729,
47449622,
63378772
],
"milestone_updates": [
484,
724,
968
],
"warmup_target_tokens": 3802726,
"learning_rate": 0.0001,
"row_lr": 0.0001,
"optimizer": "adamw",
"optimizer_betas": [
0.9,
0.95
],
"optimizer_eps": 1e-08,
"lora_rank": 64,
"lora_alpha": 128,
"lora_dropout": 0.0,
"lora_target_modules": [
"q_proj",
"k_proj",
"v_proj",
"o_proj",
"gate_proj",
"up_proj",
"down_proj"
],
"lora_weight_decay": 0.1,
"row_weight_decay": 0.0,
"max_grad_norm": 1.0,
"tokens_per_optimizer_update": 65536,
"packing": false,
"truncation": false,
"repeat_examples": false,
"precision": "bfloat16",
"trainable_dtype": "float32",
"gradient_checkpointing": true,
"eos_weight": 1.0,
"scheduler": "cosine_by_assistant_target_token_dose",
"worker_lr_scale": "rstar_dual_worker_v4._lr_scale",
"supervised_tail": [
100257
],
"trainable_token_indices": {
"embed_tokens": [
100257
],
"lm_head": [
100257
]
},
"untie_tied_embeddings": true,
"context": 32768,
"eval": {
"suite": "DEV256",
"seed": 3407,
"temperature": 0.6,
"top_p": 0.95,
"top_k": 20,
"ctx": 32768,
"stop_ids": [
100257,
100265
],
"base_reeval_same_mode": true
}
},
"code": {
"path": "/workspace/tools/rstar_dual_worker_v4.py",
"commit_or_sha256": "97a298e9ee4f0bb571baa88c7d75285ade2ea2a8a9b5875b35db25b5db9b3842",
"template": "/workspace/code-sft-runs/t30b2507-q4b-tn-v2-lr1e4-e1-20260907/templates/O7B-THINK.template_v3.json"
},
"started_at": "2026-09-08T07:50:11Z",
"endpoint": "2EP_ONLY @63378772 assistant tokens; no checkpoint picking",
"status": "RUNNING"
}