Hugging Face's logo

helblazer811
/
FlowReasoningModel

PyTorch
frm
flow-reasoning-models
reasoning
Model card Files
xet
 Community
FlowReasoningModel
File size: 3,238 Bytes
3e8c45f
 
f94da20
 
3e8c45f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
f94da20
 
 
3e8c45f
 
 
 
 
f94da20
3e8c45f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
f94da20
3e8c45f
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
{
  "dataset": "zebra",
  "checkpoint": "/proj/grp-flow-map/LoopedFlows/experiments/public_simplified_repro/outputs/runs/20261015-1119/full/zebra_s0/stage_b/best.pt",
  "checkpoint_step": 115000,
  "use_ema": true,
  "protocol": {
    "split": "test",
    "num_samples": 1024,
    "subset_seed": 0,
    "seed": 0,
    "batch_size": 64,
    "sampler": "sde",
    "steps": 256,
    "noise_scale": 0.2,
    "held_level": 0.4,
    "held_iters": 1023,
    "renoise": 1.0,
    "self_cond": true
  },
  "metrics": {
    "n_samples": 1024,
    "exact_match": 0.99609375,
    "solved": 0.99609375,
    "token_accuracy": 0.9985546875,
    "clue_consistency": 1.0
  },
  "checkpoint_provenance": {
    "dataset": "zebra",
    "stage": "b",
    "seed": 0,
    "recipe": {
      "data": {
        "dataset": "zebra",
        "root": "/proj/grp-flow-map/FlowMapReasoning/datasets",
        "train_max_samples": null,
        "balanced_by_rating": true,
        "symmetry_probability": 0.0,
        "sizes": [
          "5x5"
        ],
        "repack": true,
        "pin_answer": false
      },
      "model": {
        "hidden_size": 384,
        "cond_dim": 128,
        "length": 81,
        "n_blocks": 12,
        "n_heads": 12,
        "dropout": 0.1,
        "self_cond": true
      },
      "train": {
        "method": "flm",
        "seq_len": 81,
        "vocab_size": 10,
        "lr": 0.0003,
        "weight_decay": 0.0,
        "grad_clip": 1.0,
        "betas": [
          0.9,
          0.999
        ],
        "eps": 1e-08,
        "ema_decay": 0.9999,
        "fused_adamw": true,
        "t_min": 0.0,
        "t_max": 1.0,
        "max_steps": 100000,
        "local_batch_size": 128,
        "global_batch_size": 0,
        "warmup_steps": 0,
        "lr_schedule": "constant",
        "lr_min_ratio": 0.1,
        "log_every": 100,
        "val_every": 5000,
        "save_best_metric": "val/solved",
        "checkpoint_dir": "checkpoints",
        "save_every": 5000,
        "resume": null,
        "keep_last_only": false,
        "reset_lr_on_resume": false,
        "no_wandb": true,
        "wandb_project": "flowmap",
        "wandb_run_name": null
      },
      "loss": {
        "self_cond_prob": 0.5,
        "self_forcing_rollout": 0.5,
        "sfr_t_min": 0.0,
        "sfr_fixed_depth": 16,
        "sfr_deep_sup_n": 4
      },
      "evaluation": {
        "split": "test",
        "num_samples": 1024,
        "subset_seed": 0,
        "seed": 0,
        "batch_size": 64,
        "sampler": "sde",
        "steps": 256,
        "noise_scale": 0.2,
        "held_level": 0.4,
        "held_iters": 1023,
        "renoise": 1.0,
        "self_cond": true
      },
      "validation": {
        "split": "test",
        "num_samples": 256,
        "subset_seed": 0,
        "seed": 0,
        "batch_size": 256,
        "sampler": "euler",
        "steps": 128,
        "noise_scale": 0.0,
        "held_level": 0.4,
        "held_iters": 1023,
        "renoise": 1.0,
        "self_cond": true
      },
      "pipeline": {
        "seed": 0,
        "num_workers": 2,
        "stage_a_steps": 200000,
        "stage_b_steps": 200000,
        "warm_start": "best"
      }
    }
  },
  "nfe_per_puzzle": 257
}