Instructions to use interfaceinde/laya-gate with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Laya
How to use interfaceinde/laya-gate with Laya:
# No code snippets available yet for this library. # To use this model, check the repository files and the library's documentation. # Want to help? PRs adding snippets are welcome at: # https://github.com/huggingface/huggingface.js
- Notebooks
- Google Colab
- Kaggle
File size: 4,130 Bytes
8af56bb | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 | {
"eval_source": "data\\sets\\val.jsonl",
"eval_mode": "overlapping_eval",
"is_held_out": false,
"eval_items": 3118,
"note": "Evaluation set contains rows that overlap training data (8/3118 items); metrics do not reflect strictly independent held-out evaluation.",
"before": {
"items": 3118,
"loss": 2.4687,
"accuracy": 0.4641,
"mean_confidence": 0.8268,
"ece": 0.3628,
"brier": 0.8545,
"brier_top1": 0.3896,
"by_type": {
"choice": {
"items": 1164,
"loss": 1.7198,
"accuracy": 0.5112,
"mean_confidence": 0.7506,
"ece": 0.2394,
"brier": 0.7207,
"brier_top1": 0.2891
},
"score": {
"items": 374,
"loss": 2.0527,
"accuracy": 0.2941,
"mean_confidence": 0.7141,
"ece": 0.42,
"brier": 0.9597,
"brier_top1": 0.3878
},
"noul": {
"items": 1580,
"loss": 3.1189,
"accuracy": 0.4696,
"mean_confidence": 0.9097,
"ece": 0.4401,
"brier": 0.9281,
"brier_top1": 0.464
}
}
},
"after": {
"items": 3118,
"loss": 0.2564,
"accuracy": 0.9622,
"mean_confidence": 0.9868,
"ece": 0.0263,
"brier": 0.0724,
"brier_top1": 0.0355,
"by_type": {
"choice": {
"items": 1164,
"loss": 0.2698,
"accuracy": 0.9424,
"mean_confidence": 0.972,
"ece": 0.0309,
"brier": 0.1056,
"brier_top1": 0.051
},
"score": {
"items": 374,
"loss": 0.8551,
"accuracy": 0.9465,
"mean_confidence": 0.9953,
"ece": 0.0581,
"brier": 0.1106,
"brier_top1": 0.0553
},
"noul": {
"items": 1580,
"loss": 0.1047,
"accuracy": 0.9804,
"mean_confidence": 0.9958,
"ece": 0.016,
"brier": 0.0388,
"brier_top1": 0.0194
}
}
},
"comparison": {
"delta_accuracy": 0.4981,
"delta_loss": -2.2123,
"delta_ece": -0.3365,
"delta_brier": -0.7821,
"delta_brier_top1": -0.3541,
"delta_mean_confidence": 0.16
},
"calibration": {
"choice": {
"items": 139,
"temperature": 3.7832961082458496,
"issues": [],
"accuracy": 0.9712230215827338,
"mean_confidence": 0.9759824520505122
},
"score": {
"items": 44,
"temperature": 1.0,
"issues": [
"fitted on only 44 calibration items",
"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": 0.999999929841763
},
"noul": {
"items": 217,
"temperature": 3.33156156539917,
"issues": [],
"accuracy": 0.9953917050691244,
"mean_confidence": 0.9961188719812915
}
},
"abstention_thresholds": {
"score:3-5": 1.0,
"noul:2": 0.9763,
"choice:3-5": 0.9681,
"choice:2": 0.9901,
"choice:6-10": 0.8453
},
"training": {
"epochs": 5,
"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": "rlcd",
"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": 2048,
"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": "data\\sets\\val.jsonl",
"target_error": 0.1,
"min_abstain_n": 10
}
} |