206 lines
3.5 KiB
JSON
206 lines
3.5 KiB
JSON
{
|
|
"mode": "rand",
|
|
"seed": 42,
|
|
"hf_repo": "Likithp/v8_rand_s42",
|
|
"base_model": "Qwen/Qwen2.5-0.5B-Instruct",
|
|
"dataset": "data/cs8_rand_v4",
|
|
"hf_dataset": "Likithp/cs8_rand_v4",
|
|
"trained_at": "2026-06-04T04:26:06.349629+00:00",
|
|
"optimizer": "AdamW",
|
|
"lr": 3e-05,
|
|
"weight_decay": 0.0,
|
|
"batch_size": 8,
|
|
"grad_accum": 8,
|
|
"effective_batch": 64,
|
|
"epochs": 5,
|
|
"warmup_steps": 200,
|
|
"grad_clip": 0.5,
|
|
"lr_schedule": "cosine",
|
|
"max_seq_len": 256,
|
|
"dtype": "bfloat16",
|
|
"alias_groups": {
|
|
"T1": [
|
|
"emp",
|
|
"inv",
|
|
"txn",
|
|
"mgr",
|
|
"ord",
|
|
"prod",
|
|
"cust",
|
|
"dept",
|
|
"acct",
|
|
"sale"
|
|
],
|
|
"T2": [
|
|
"tsk",
|
|
"brc",
|
|
"rgn",
|
|
"lne",
|
|
"prj",
|
|
"empl",
|
|
"ordr",
|
|
"catg",
|
|
"schd",
|
|
"shp"
|
|
]
|
|
},
|
|
"log_every_steps": 50,
|
|
"eval_n": 500,
|
|
"dry_run": false,
|
|
"best_epoch": 5,
|
|
"val_exact_match": 0.036,
|
|
"val_outer_alias_acc": 1.0,
|
|
"val_inner_alias_acc": 0.039,
|
|
"val_t1_outer_alias_acc": 1.0,
|
|
"val_t2_outer_alias_acc": 1.0,
|
|
"val_t1_t2_gap_pp": 0.0,
|
|
"val_inner_by_alias": {
|
|
"acct": {
|
|
"correct": 0,
|
|
"total": 66,
|
|
"pct": 0.0,
|
|
"token_group": "T1"
|
|
},
|
|
"brc": {
|
|
"correct": 0,
|
|
"total": 42,
|
|
"pct": 0.0,
|
|
"token_group": "T2"
|
|
},
|
|
"catg": {
|
|
"correct": 1,
|
|
"total": 47,
|
|
"pct": 2.1,
|
|
"token_group": "T2"
|
|
},
|
|
"cust": {
|
|
"correct": 3,
|
|
"total": 52,
|
|
"pct": 5.8,
|
|
"token_group": "T1"
|
|
},
|
|
"dept": {
|
|
"correct": 0,
|
|
"total": 57,
|
|
"pct": 0.0,
|
|
"token_group": "T1"
|
|
},
|
|
"emp": {
|
|
"correct": 0,
|
|
"total": 46,
|
|
"pct": 0.0,
|
|
"token_group": "T1"
|
|
},
|
|
"empl": {
|
|
"correct": 2,
|
|
"total": 36,
|
|
"pct": 5.6,
|
|
"token_group": "T2"
|
|
},
|
|
"inv": {
|
|
"correct": 8,
|
|
"total": 56,
|
|
"pct": 14.3,
|
|
"token_group": "T1"
|
|
},
|
|
"lne": {
|
|
"correct": 1,
|
|
"total": 38,
|
|
"pct": 2.6,
|
|
"token_group": "T2"
|
|
},
|
|
"mgr": {
|
|
"correct": 2,
|
|
"total": 48,
|
|
"pct": 4.2,
|
|
"token_group": "T1"
|
|
},
|
|
"ord": {
|
|
"correct": 2,
|
|
"total": 47,
|
|
"pct": 4.3,
|
|
"token_group": "T1"
|
|
},
|
|
"ordr": {
|
|
"correct": 1,
|
|
"total": 53,
|
|
"pct": 1.9,
|
|
"token_group": "T2"
|
|
},
|
|
"prj": {
|
|
"correct": 2,
|
|
"total": 51,
|
|
"pct": 3.9,
|
|
"token_group": "T2"
|
|
},
|
|
"prod": {
|
|
"correct": 0,
|
|
"total": 52,
|
|
"pct": 0.0,
|
|
"token_group": "T1"
|
|
},
|
|
"rgn": {
|
|
"correct": 1,
|
|
"total": 56,
|
|
"pct": 1.8,
|
|
"token_group": "T2"
|
|
},
|
|
"sale": {
|
|
"correct": 3,
|
|
"total": 46,
|
|
"pct": 6.5,
|
|
"token_group": "T1"
|
|
},
|
|
"schd": {
|
|
"correct": 3,
|
|
"total": 62,
|
|
"pct": 4.8,
|
|
"token_group": "T2"
|
|
},
|
|
"shp": {
|
|
"correct": 8,
|
|
"total": 52,
|
|
"pct": 15.4,
|
|
"token_group": "T2"
|
|
},
|
|
"tsk": {
|
|
"correct": 2,
|
|
"total": 44,
|
|
"pct": 4.5,
|
|
"token_group": "T2"
|
|
},
|
|
"txn": {
|
|
"correct": 0,
|
|
"total": 49,
|
|
"pct": 0.0,
|
|
"token_group": "T1"
|
|
}
|
|
},
|
|
"train_log": [
|
|
{
|
|
"epoch": 1,
|
|
"loss": 0.264341,
|
|
"val_em": 0.06
|
|
},
|
|
{
|
|
"epoch": 2,
|
|
"loss": 0.096242,
|
|
"val_em": 0.046
|
|
},
|
|
{
|
|
"epoch": 3,
|
|
"loss": 0.094966,
|
|
"val_em": 0.05
|
|
},
|
|
{
|
|
"epoch": 4,
|
|
"loss": 0.093876,
|
|
"val_em": 0.048
|
|
},
|
|
{
|
|
"epoch": 5,
|
|
"loss": 0.093402,
|
|
"val_em": 0.034
|
|
}
|
|
]
|
|
} |