File size: 10,932 Bytes
c7ebaa1
 
 
 
 
 
 
 
 
 
 
 
 
bff2f94
c7ebaa1
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
bff2f94
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
c7ebaa1
bff2f94
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
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
"""
Command-line interface for BioRLHF.

This module provides CLI entry points for training and evaluating models.
"""

import argparse
import json
import sys
from pathlib import Path

from biorlhf.training.sft import SFTTrainingConfig, run_sft_training
from biorlhf.evaluation.evaluate import evaluate_model as _evaluate_model
from biorlhf.training.grpo import BioGRPOConfig, run_grpo_training


def train():
    """CLI entry point for training models."""
    parser = argparse.ArgumentParser(
        description="Train a BioRLHF model using supervised fine-tuning",
        formatter_class=argparse.ArgumentDefaultsHelpFormatter,
    )

    # Model settings
    parser.add_argument(
        "--model",
        type=str,
        default="mistralai/Mistral-7B-v0.3",
        help="Base model to fine-tune",
    )
    parser.add_argument(
        "--dataset",
        type=str,
        required=True,
        help="Path to training dataset JSON file",
    )
    parser.add_argument(
        "--output",
        type=str,
        default="./biorlhf_model",
        help="Output directory for trained model",
    )

    # Training hyperparameters
    parser.add_argument(
        "--epochs",
        type=int,
        default=3,
        help="Number of training epochs",
    )
    parser.add_argument(
        "--batch-size",
        type=int,
        default=4,
        help="Training batch size per device",
    )
    parser.add_argument(
        "--learning-rate",
        type=float,
        default=2e-4,
        help="Learning rate",
    )
    parser.add_argument(
        "--max-length",
        type=int,
        default=1024,
        help="Maximum sequence length",
    )

    # LoRA settings
    parser.add_argument(
        "--lora-r",
        type=int,
        default=64,
        help="LoRA rank",
    )
    parser.add_argument(
        "--lora-alpha",
        type=int,
        default=128,
        help="LoRA alpha",
    )

    # Other settings
    parser.add_argument(
        "--no-quantization",
        action="store_true",
        help="Disable 4-bit quantization",
    )
    parser.add_argument(
        "--no-wandb",
        action="store_true",
        help="Disable Weights & Biases logging",
    )
    parser.add_argument(
        "--wandb-project",
        type=str,
        default="biorlhf",
        help="W&B project name",
    )
    parser.add_argument(
        "--wandb-run-name",
        type=str,
        default="sft_training",
        help="W&B run name",
    )
    parser.add_argument(
        "--config",
        type=str,
        default=None,
        help="Path to JSON config file (overrides other args)",
    )

    args = parser.parse_args()

    # Validate dataset path
    if not Path(args.dataset).exists():
        print(f"Error: Dataset not found at {args.dataset}", file=sys.stderr)
        sys.exit(1)

    # Load config from file if provided
    if args.config:
        with open(args.config) as f:
            config_dict = json.load(f)
        config = SFTTrainingConfig(**config_dict)
    else:
        config = SFTTrainingConfig(
            model_name=args.model,
            dataset_path=args.dataset,
            output_dir=args.output,
            num_epochs=args.epochs,
            batch_size=args.batch_size,
            learning_rate=args.learning_rate,
            max_length=args.max_length,
            lora_r=args.lora_r,
            lora_alpha=args.lora_alpha,
            use_4bit=not args.no_quantization,
            use_wandb=not args.no_wandb,
            wandb_project=args.wandb_project,
            wandb_run_name=args.wandb_run_name,
        )

    print("BioRLHF Training")
    print("=" * 50)
    print(f"Model: {config.model_name}")
    print(f"Dataset: {config.dataset_path}")
    print(f"Output: {config.output_dir}")
    print("=" * 50)

    try:
        output_path = run_sft_training(config)
        print(f"\nModel saved to: {output_path}")
    except Exception as e:
        print(f"Error during training: {e}", file=sys.stderr)
        sys.exit(1)


def evaluate():
    """CLI entry point for evaluating models."""
    parser = argparse.ArgumentParser(
        description="Evaluate a BioRLHF model on a test set",
        formatter_class=argparse.ArgumentDefaultsHelpFormatter,
    )

    parser.add_argument(
        "--model",
        type=str,
        required=True,
        help="Path to the fine-tuned model directory",
    )
    parser.add_argument(
        "--test-set",
        type=str,
        required=True,
        help="Path to test questions JSON file",
    )
    parser.add_argument(
        "--base-model",
        type=str,
        default="mistralai/Mistral-7B-v0.3",
        help="Base model name",
    )
    parser.add_argument(
        "--output",
        type=str,
        default=None,
        help="Output path for detailed results JSON",
    )
    parser.add_argument(
        "--no-quantization",
        action="store_true",
        help="Disable 4-bit quantization",
    )
    parser.add_argument(
        "--temperature",
        type=float,
        default=0.1,
        help="Generation temperature (0 for greedy)",
    )
    parser.add_argument(
        "--max-tokens",
        type=int,
        default=512,
        help="Maximum tokens to generate",
    )

    args = parser.parse_args()

    # Validate paths
    if not Path(args.model).exists():
        print(f"Error: Model not found at {args.model}", file=sys.stderr)
        sys.exit(1)

    if not Path(args.test_set).exists():
        print(f"Error: Test set not found at {args.test_set}", file=sys.stderr)
        sys.exit(1)

    print("BioRLHF Evaluation")
    print("=" * 50)
    print(f"Model: {args.model}")
    print(f"Test Set: {args.test_set}")
    print("=" * 50)

    try:
        results = _evaluate_model(
            model_path=args.model,
            test_questions_path=args.test_set,
            base_model=args.base_model,
            use_4bit=not args.no_quantization,
            max_new_tokens=args.max_tokens,
            temperature=args.temperature,
        )

        print("\nResults:")
        print("-" * 30)
        print(f"Overall Accuracy: {results.overall_accuracy:.1%}")
        print(f"Factual Accuracy: {results.factual_accuracy:.1%}")
        print(f"Reasoning Accuracy: {results.reasoning_accuracy:.1%}")
        print(f"Calibration Accuracy: {results.calibration_accuracy:.1%}")
        print(f"Total: {results.correct_answers}/{results.total_questions}")

        # Save detailed results if requested
        if args.output:
            output_data = {
                "model_path": args.model,
                "test_set": args.test_set,
                "metrics": {
                    "overall_accuracy": results.overall_accuracy,
                    "factual_accuracy": results.factual_accuracy,
                    "reasoning_accuracy": results.reasoning_accuracy,
                    "calibration_accuracy": results.calibration_accuracy,
                    "total_questions": results.total_questions,
                    "correct_answers": results.correct_answers,
                },
                "detailed_results": results.detailed_results,
            }

            with open(args.output, "w") as f:
                json.dump(output_data, f, indent=2)

            print(f"\nDetailed results saved to: {args.output}")

    except Exception as e:
        print(f"Error during evaluation: {e}", file=sys.stderr)
        sys.exit(1)


def grpo_train():
    """CLI entry point for GRPO training with biological verifiers."""
    parser = argparse.ArgumentParser(
        description="Train a BioGRPO model with composable biological verifiers",
        formatter_class=argparse.ArgumentDefaultsHelpFormatter,
    )

    parser.add_argument(
        "--model",
        type=str,
        default="mistralai/Mistral-7B-v0.3",
        help="Base model to fine-tune",
    )
    parser.add_argument(
        "--sft-model",
        type=str,
        default=None,
        help="Path to SFT checkpoint (recommended)",
    )
    parser.add_argument(
        "--output",
        type=str,
        default="./biogrpo_model",
        help="Output directory",
    )
    parser.add_argument(
        "--num-generations",
        type=int,
        default=8,
        help="G value: number of completions per prompt",
    )
    parser.add_argument(
        "--beta",
        type=float,
        default=0.04,
        help="KL penalty coefficient",
    )
    parser.add_argument(
        "--learning-rate",
        type=float,
        default=1e-6,
        help="Learning rate",
    )
    parser.add_argument(
        "--lora-r",
        type=int,
        default=32,
        help="LoRA rank",
    )
    parser.add_argument(
        "--lora-alpha",
        type=int,
        default=64,
        help="LoRA alpha",
    )
    parser.add_argument(
        "--verifiers",
        type=str,
        nargs="+",
        default=None,
        help="Active verifiers (e.g., V1 V2 V3 V4). Default: all",
    )
    parser.add_argument(
        "--pathway-db",
        type=str,
        default="hallmark",
        choices=["hallmark", "kegg", "reactome", "mitocarta"],
        help="Pathway database for GeneLab questions",
    )
    parser.add_argument(
        "--no-wandb",
        action="store_true",
        help="Disable W&B logging",
    )
    parser.add_argument(
        "--wandb-project",
        type=str,
        default="biogrpo",
        help="W&B project name",
    )
    parser.add_argument(
        "--wandb-run-name",
        type=str,
        default="grpo_v1",
        help="W&B run name",
    )
    parser.add_argument(
        "--config",
        type=str,
        default=None,
        help="Path to JSON config file (overrides other args)",
    )

    args = parser.parse_args()

    if args.config:
        with open(args.config) as f:
            config_dict = json.load(f)
        config = BioGRPOConfig(**config_dict)
    else:
        config = BioGRPOConfig(
            model_name=args.model,
            sft_model_path=args.sft_model,
            output_dir=args.output,
            num_generations=args.num_generations,
            beta=args.beta,
            learning_rate=args.learning_rate,
            lora_r=args.lora_r,
            lora_alpha=args.lora_alpha,
            active_verifiers=args.verifiers,
            pathway_db=args.pathway_db,
            use_wandb=not args.no_wandb,
            wandb_project=args.wandb_project,
            wandb_run_name=args.wandb_run_name,
        )

    try:
        output_path = run_grpo_training(config)
        print(f"\nModel saved to: {output_path}")
    except Exception as e:
        import traceback
        traceback.print_exc()
        print(f"Error during GRPO training: {e}", file=sys.stderr)
        sys.exit(1)


if __name__ == "__main__":
    print("Use 'biorlhf-train', 'biorlhf-evaluate', or 'biorlhf-grpo' commands after installation.")