196 lines
3.4 KiB
JSON
196 lines
3.4 KiB
JSON
{
|
|
"model_size": "0.5B",
|
|
"mode": "fixed",
|
|
"seed": 42,
|
|
"hf_repo": "Likithp/sensor_v1_0.5B_fixed_s42",
|
|
"base_model": "Qwen/Qwen2.5-0.5B",
|
|
"dataset": "data/sensor_fixed_v1",
|
|
"dataset_version": "sensor_v1",
|
|
"trained_at": "2026-07-01T02:08:07.846344+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",
|
|
"checkpoint_criterion": "val_em",
|
|
"max_seq_len": 256,
|
|
"dtype": "torch.bfloat16",
|
|
"eval_method": "teacher_forcing_argmax",
|
|
"alias_groups": {
|
|
"T1": [
|
|
"thr",
|
|
"gst",
|
|
"prt",
|
|
"clr"
|
|
],
|
|
"T2": [
|
|
"rnfl",
|
|
"hgr",
|
|
"grvl",
|
|
"brt",
|
|
"mbl",
|
|
"nbl",
|
|
"evp",
|
|
"cnd",
|
|
"frs",
|
|
"snw",
|
|
"flx",
|
|
"tnd",
|
|
"drt",
|
|
"ftu"
|
|
]
|
|
},
|
|
"log_every_steps": 50,
|
|
"eval_n": 500,
|
|
"dry_run": false,
|
|
"best_epoch": 5,
|
|
"best_train_loss": null,
|
|
"val_exact_match": 1.0,
|
|
"val_outer_alias_acc": 1.0,
|
|
"val_inner_alias_acc": 1.0,
|
|
"val_t1_inner_alias_acc": 1.0,
|
|
"val_t2_inner_alias_acc": 1.0,
|
|
"val_t1_t2_gap_pp": 0.0,
|
|
"val_inner_by_alias": {
|
|
"brt": {
|
|
"correct": 67,
|
|
"total": 67,
|
|
"pct": 100.0,
|
|
"token_group": "T2"
|
|
},
|
|
"clr": {
|
|
"correct": 61,
|
|
"total": 61,
|
|
"pct": 100.0,
|
|
"token_group": "T1"
|
|
},
|
|
"cnd": {
|
|
"correct": 60,
|
|
"total": 60,
|
|
"pct": 100.0,
|
|
"token_group": "T2"
|
|
},
|
|
"drt": {
|
|
"correct": 59,
|
|
"total": 59,
|
|
"pct": 100.0,
|
|
"token_group": "T2"
|
|
},
|
|
"evp": {
|
|
"correct": 70,
|
|
"total": 70,
|
|
"pct": 100.0,
|
|
"token_group": "T2"
|
|
},
|
|
"flx": {
|
|
"correct": 51,
|
|
"total": 51,
|
|
"pct": 100.0,
|
|
"token_group": "T2"
|
|
},
|
|
"frs": {
|
|
"correct": 59,
|
|
"total": 59,
|
|
"pct": 100.0,
|
|
"token_group": "T2"
|
|
},
|
|
"ftu": {
|
|
"correct": 59,
|
|
"total": 59,
|
|
"pct": 100.0,
|
|
"token_group": "T2"
|
|
},
|
|
"grvl": {
|
|
"correct": 45,
|
|
"total": 45,
|
|
"pct": 100.0,
|
|
"token_group": "T2"
|
|
},
|
|
"gst": {
|
|
"correct": 50,
|
|
"total": 50,
|
|
"pct": 100.0,
|
|
"token_group": "T1"
|
|
},
|
|
"hgr": {
|
|
"correct": 51,
|
|
"total": 51,
|
|
"pct": 100.0,
|
|
"token_group": "T2"
|
|
},
|
|
"mbl": {
|
|
"correct": 57,
|
|
"total": 57,
|
|
"pct": 100.0,
|
|
"token_group": "T2"
|
|
},
|
|
"nbl": {
|
|
"correct": 52,
|
|
"total": 52,
|
|
"pct": 100.0,
|
|
"token_group": "T2"
|
|
},
|
|
"prt": {
|
|
"correct": 45,
|
|
"total": 45,
|
|
"pct": 100.0,
|
|
"token_group": "T1"
|
|
},
|
|
"rnfl": {
|
|
"correct": 56,
|
|
"total": 56,
|
|
"pct": 100.0,
|
|
"token_group": "T2"
|
|
},
|
|
"snw": {
|
|
"correct": 44,
|
|
"total": 44,
|
|
"pct": 100.0,
|
|
"token_group": "T2"
|
|
},
|
|
"thr": {
|
|
"correct": 64,
|
|
"total": 64,
|
|
"pct": 100.0,
|
|
"token_group": "T1"
|
|
},
|
|
"tnd": {
|
|
"correct": 50,
|
|
"total": 50,
|
|
"pct": 100.0,
|
|
"token_group": "T2"
|
|
}
|
|
},
|
|
"train_log": [
|
|
{
|
|
"epoch": 1,
|
|
"loss": 0.064943,
|
|
"val_em": 1.0
|
|
},
|
|
{
|
|
"epoch": 2,
|
|
"loss": 1e-05,
|
|
"val_em": 1.0
|
|
},
|
|
{
|
|
"epoch": 3,
|
|
"loss": 1e-05,
|
|
"val_em": 1.0
|
|
},
|
|
{
|
|
"epoch": 4,
|
|
"loss": 1e-05,
|
|
"val_em": 1.0
|
|
},
|
|
{
|
|
"epoch": 5,
|
|
"loss": 1e-05,
|
|
"val_em": 1.0
|
|
}
|
|
]
|
|
} |