Hugging Face's logo

trenify
/
jerzyk-1.0

Text Classification
Laya
Safetensors
Polish
decision-model
systemone
jev
polish
legal
document-classification
calibrated
Eval Results (legacy)
Model card Files
xet
 Community
jerzyk-1.0
File size: 3,836 Bytes
ef69c3e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
{
  "eval_source": "calibration_slice",
  "eval_mode": "in_sample_calibration",
  "is_held_out": false,
  "eval_items": 400,
  "note": "Metrics reflect in-sample calibration fit, not independent held-out evaluation. Provide --eval for held-out validation.",
  "before": {
    "items": 400,
    "loss": 1.5651,
    "accuracy": 0.56,
    "mean_confidence": 0.8108,
    "ece": 0.2508,
    "brier": 0.6589,
    "brier_top1": 0.2824,
    "by_type": {
      "choice": {
        "items": 136,
        "loss": 1.2561,
        "accuracy": 0.7868,
        "mean_confidence": 0.898,
        "ece": 0.1288,
        "brier": 0.3576,
        "brier_top1": 0.1483
      },
      "score": {
        "items": 129,
        "loss": 2.1186,
        "accuracy": 0.3101,
        "mean_confidence": 0.6873,
        "ece": 0.3772,
        "brier": 0.9646,
        "brier_top1": 0.3685
      },
      "noul": {
        "items": 135,
        "loss": 1.3475,
        "accuracy": 0.5704,
        "mean_confidence": 0.8411,
        "ece": 0.2768,
        "brier": 0.6702,
        "brier_top1": 0.3351
      }
    }
  },
  "after": {
    "items": 400,
    "loss": 0.0509,
    "accuracy": 0.985,
    "mean_confidence": 0.9841,
    "ece": 0.0104,
    "brier": 0.0242,
    "brier_top1": 0.0119,
    "by_type": {
      "choice": {
        "items": 136,
        "loss": 0.0,
        "accuracy": 1.0,
        "mean_confidence": 1.0,
        "ece": 0.0,
        "brier": 0.0,
        "brier_top1": 0.0
      },
      "score": {
        "items": 129,
        "loss": 0.0468,
        "accuracy": 0.9845,
        "mean_confidence": 0.985,
        "ece": 0.0107,
        "brier": 0.0247,
        "brier_top1": 0.0119
      },
      "noul": {
        "items": 135,
        "loss": 0.1061,
        "accuracy": 0.9704,
        "mean_confidence": 0.9671,
        "ece": 0.0218,
        "brier": 0.048,
        "brier_top1": 0.024
      }
    }
  },
  "comparison": {
    "delta_accuracy": 0.425,
    "delta_loss": -1.5142,
    "delta_ece": -0.2404,
    "delta_brier": -0.6347,
    "delta_brier_top1": -0.2705,
    "delta_mean_confidence": 0.1733
  },
  "calibration": {
    "choice": {
      "items": 136,
      "temperature": 0.9999958276748657,
      "issues": [
        "the fit came back unchanged at 1.0 (accuracy 1.00, mean confidence 1.000 on the calibration items), which usually means they gave it nothing to correct; this type's confidences are uncalibrated"
      ],
      "accuracy": 1.0,
      "mean_confidence": 0.9999959610103946
    },
    "score": {
      "items": 129,
      "temperature": 1.8285092115402222,
      "issues": [],
      "accuracy": 0.9844961240310077,
      "mean_confidence": 0.9849781946721331
    },
    "noul": {
      "items": 135,
      "temperature": 3.370203733444214,
      "issues": [],
      "accuracy": 0.9703703703703703,
      "mean_confidence": 0.9671178420842376
    }
  },
  "abstention_thresholds": {
    "choice:6-10": 0.9998,
    "score:3-5": 0.5209,
    "noul:2": 0.5493
  },
  "training": {
    "epochs": 4,
    "micro_batch": 2,
    "grad_accum": 8,
    "encoder_lr": 2.5e-05,
    "head_lr": 0.0001,
    "min_lr": 1e-06,
    "weight_decay": 0.01,
    "grad_clip": 1.0,
    "loss": "soft-ce",
    "rl_samples": 4,
    "sigma_start": 0.4,
    "sigma_end": 0.1,
    "w_sph": 0.75,
    "w_rps": 1.0,
    "shuffle_options": [
      "choice"
    ],
    "option_layout": null,
    "freeze_encoder": false,
    "calib_max": 400,
    "calib_frac": 0.1,
    "calib_seed": 20260922,
    "seed": 0,
    "max_len": null,
    "head_max_len": null,
    "amp": null,
    "gradient_checkpointing": null,
    "log_every": 100,
    "label_smoothing": 0.0,
    "text_column": "text",
    "label_column": "label",
    "question_id": "label",
    "instructions": null,
    "eval_data": null,
    "target_error": 0.1,
    "min_abstain_n": 10
  }
}