{ "adapter_path": "runs/qwen3_14b_restricted_score_ce/adapter", "allowed_scores": [ "1", "2", "3", "4", "5" ], "artifacts": { "config.yaml": { "data": { "columns": { "benchmark": "benchmark", "example_id": "instance_id", "judge_1_score": "score_gpt4o", "judge_2_score": "score_sonnet", "model_id": "model_id", "question": "question", "reference_answer": "ground_truth", "response": "response", "source": "source", "task": "task" }, "filters": { "max_disagreement": 1, "max_response_tokens": 4096, "on_sequence_overflow": "skip" }, "path": "data/processed/response_scores.parquet", "prepared_dir": null, "preprocessing_num_workers": 40, "token_length_batch_size": 2048, "tokenizers_parallelism": true }, "distributed": { "backend": "nccl", "deepspeed": { "config_overrides": {}, "gradient_clipping": "auto", "offload_optimizer_device": "none", "offload_param_device": "none", "stage3_gather_16bit_weights_on_model_save": true, "zero_stage": 2 }, "enabled": true, "find_unused_parameters": false, "fsdp": { "activation_checkpointing": true, "sharding_strategy": "full_shard", "transformer_layer_cls_to_wrap": null, "use_orig_params": true }, "gradient_checkpointing": false, "mixed_precision": "bf16", "strategy": "ddp" }, "evaluation": { "length_buckets": [ 0, 256, 512, 1024, 2048, 3072, 4096, 1000000000 ] }, "hub": { "commit_message": "Upload ADeLe distilled judge", "create_pr": false, "local_checkpoint_dir": null, "max_shard_size": "5GB", "output_staging_dir": null, "private": false, "repo_id": null }, "inference": { "allow_base_model": true, "allowed_scores": [ "1", "2", "3", "4", "5" ], "batch_size": 64, "binary_threshold": 3, "generation_fallback": false, "method": "restricted_continuation_logprobs_fast", "require_adapter": false }, "model": { "adapter_path": null, "attn_implementation": "sdpa", "model_name_or_path": "Qwen/Qwen3-14B", "revision": null, "thinking_mode": { "apply_if_supported": true, "enabled": false }, "trust_remote_code": true }, "project": { "output_dir": "runs/qwen3_14b_restricted_score_ce", "run_name": "qwen3_14b_restricted_score_ce", "seed": 42 }, "prompt": { "system_prompt": "You are an answer quality evaluator. You will receive a question, a reference\nanswer, and a model response, each delimited by ### headers.\n\nRate the model response on the following scale:\n1 - Surely incorrect: the response is wrong or completely misses the point.\n2 - Likely incorrect: the response has significant errors or critical omissions.\n3 - Minimally correct / sufficient: the response is partially correct or borderline.\n4 - Likely correct: the response is mostly correct with minor issues.\n5 - Surely correct: the response is fully correct and complete.\n\nRules:\n- Base your rating strictly on correctness relative to the reference answer.\n- Ignore writing style, verbosity, or formatting unless they affect correctness.\n- If the final answer is correct but the reasoning contains substantive errors, lower the rating according to the severity of those errors.\n- Output only a single integer from 1 to 5. No explanation. No punctuation.\n" }, "split": { "held_out_model": null, "lomo_validation_fraction": 0.05, "lomo_validation_max_examples": 30000, "lomo_validation_seed": 42, "mode": "fixed_by_model", "train_models": "auto_except_val_test", "validation_models": [ "gemini-3-flash", "DK-R1-Dist-Qwen-14B", "llama3d2-3b" ] }, "training": { "cache_tokenized_datasets": true, "class_weighting": null, "dtype": "bfloat16", "eval_steps": 500, "eval_subset_seed": 42, "eval_subset_size": null, "eval_subset_strategy": "stratified", "eval_subset_stratify_columns": [ "model_id", "target_score" ], "gradient_accumulation_steps": 4, "learning_rate": 3e-05, "length_column_name": "length", "load_in_4bit": true, "logging_steps": 10, "lora_alpha": 64, "lora_dropout": 0.0, "lora_r": 32, "loss": { "class_weights": null, "lambda_binary": 0.5, "type": "ce_5way" }, "lr_scheduler_type": "cosine", "max_grad_norm": 1.0, "max_seq_length": 4096, "num_train_epochs": 1, "objective": "restricted_score_ce", "optim": "adamw_8bit", "packing": false, "per_device_eval_batch_size": 2, "per_device_train_batch_size": 2, "resume_from_checkpoint": null, "save_steps": 500, "save_total_limit": 10, "score_class_weights": null, "seed": 42, "target_modules": "auto", "train_sampling_strategy": "random", "warmup_ratio": 0.03, "weight_decay": 0.0 } }, "dataset_filtering_report.json": { "after_disagreement_filter": 287811, "after_response_length_filter": 285310, "effective_prompt_budget_tokens": 0, "examples_after_sequence_filter": 285158, "examples_before_sequence_filter": 285310, "filter_stage_distributions": { "after_disagreement_filter": { "benchmark": { "ChemLLMBench": 28771, "Civil Service Examination": 7234, "Data Analysis": 617, "Date Arithmetic": 9167, "GRE & GMAT": 3737, "LSAT": 16530, "Language": 470, "MCTACO": 3734, "MMLU-Pro": 98811, "Math": 3001, "MedCalcBench": 9804, "MenatQA": 11777, "OmniMath": 29517, "Reasoning": 975, "SAT": 7645, "SciBench": 6360, "TempReason": 11519, "TimeDial": 6087, "TimeQA": 13083, "TruthQuest": 18972 }, "model_id": { "DK-R1-Dist-Qwen-1.5B": 15314, "DK-R1-Dist-Qwen-14B": 15352, "DK-R1-Dist-Qwen-32B": 14922, "DK-R1-Dist-Qwen-7B": 15215, "gemini-2.5-flash": 14468, "gemini-3-flash": 15106, "gemini-3.1-pro": 15158, "gpt-35-turbo": 15163, "gpt-5.2": 14241, "gpt4o": 16106, "llama3d1-405b": 15418, "llama3d2-11b": 15272, "llama3d2-1b": 15274, "llama3d2-3b": 15288, "llama3d2-90b": 15393, "llama4-17B-128E": 15161, "o1-mini": 15358, "o1_re=low": 15044, "o3-mini": 14558 }, "num_examples": 287811, "target_binary": { "CORRECT": 192174, "INCORRECT": 95637 }, "target_score": { "1": 89963, "2": 5674, "3": 4685, "4": 11341, "5": 176148 }, "task": { "AMPS_Hard": 1174, "AQuA-RAT": 3737, "Algebra": 5977, "Applied Mathematics": 5444, "Calculus": 535, "Chemistry": 2539, "Date Arithmetic": 9167, "Discrete Mathematics": 5475, "E": 6185, "Geometry": 5928, "I": 6651, "LSAT-AR": 3290, "LSAT-LR": 8608, "LSAT-RC": 4632, "LogiQA-en": 7234, "MCTACO": 3734, "Math": 1901, "MenatQA-Counterfactual": 2172, "MenatQA-Order": 2738, "MenatQA-Scope": 6867, "Number Theory": 5639, "Physics": 1920, "Precalculus": 519, "S": 6136, "SAT-En": 3643, "SAT-Math": 4002, "TempReason-L2": 5565, "TempReason-L3": 5954, "TimeDial": 6087, "TimeQA-explicit": 6894, "TimeQA-implicit": 6189, "biology": 8307, "business": 7478, "chemistry": 6672, "computer science": 6271, "connections": 470, "cta": 617, "date": 491, "diagnosis": 211, "dosage": 372, "economics": 7831, "engineering": 5291, "health": 7479, "history": 5520, "lab": 3165, "law": 6462, "math": 7719, "math_comp": 1394, "molecule_captioning": 2641, "molecule_design": 4313, "name_prediction": 8652, "olympiad": 433, "other": 7799, "philosophy": 7334, "physical": 3929, "physics": 6828, "psychology": 7820, "reaction_prediction": 6929, "retrosynthesis": 6236, "risk": 1353, "severity": 283, "spatial": 582, "zebra_puzzle": 393 } }, "after_response_length_filter": { "benchmark": { "ChemLLMBench": 27913, "Civil Service Examination": 7207, "Data Analysis": 617, "Date Arithmetic": 9167, "GRE & GMAT": 3733, "LSAT": 16471, "Language": 468, "MCTACO": 3733, "MMLU-Pro": 98506, "Math": 2951, "MedCalcBench": 9799, "MenatQA": 11762, "OmniMath": 28488, "Reasoning": 953, "SAT": 7645, "SciBench": 6270, "TempReason": 11519, "TimeDial": 6087, "TimeQA": 13082, "TruthQuest": 18939 }, "model_id": { "DK-R1-Dist-Qwen-1.5B": 15302, "DK-R1-Dist-Qwen-14B": 15351, "DK-R1-Dist-Qwen-32B": 12577, "DK-R1-Dist-Qwen-7B": 15128, "gemini-2.5-flash": 14426, "gemini-3-flash": 15104, "gemini-3.1-pro": 15158, "gpt-35-turbo": 15163, "gpt-5.2": 14241, "gpt4o": 16106, "llama3d1-405b": 15418, "llama3d2-11b": 15267, "llama3d2-1b": 15272, "llama3d2-3b": 15288, "llama3d2-90b": 15393, "llama4-17B-128E": 15159, "o1-mini": 15358, "o1_re=low": 15041, "o3-mini": 14558 }, "num_examples": 285310, "target_binary": { "CORRECT": 191333, "INCORRECT": 93977 }, "target_score": { "1": 88379, "2": 5598, "3": 4641, "4": 11284, "5": 175408 }, "task": { "AMPS_Hard": 1157, "AQuA-RAT": 3733, "Algebra": 5814, "Applied Mathematics": 5309, "Calculus": 516, "Chemistry": 2502, "Date Arithmetic": 9167, "Discrete Mathematics": 5235, "E": 6164, "Geometry": 5695, "I": 6647, "LSAT-AR": 3234, "LSAT-LR": 8605, "LSAT-RC": 4632, "LogiQA-en": 7207, "MCTACO": 3733, "Math": 1887, "MenatQA-Counterfactual": 2166, "MenatQA-Order": 2737, "MenatQA-Scope": 6859, "Number Theory": 5411, "Physics": 1881, "Precalculus": 508, "S": 6128, "SAT-En": 3643, "SAT-Math": 4002, "TempReason-L2": 5565, "TempReason-L3": 5954, "TimeDial": 6087, "TimeQA-explicit": 6893, "TimeQA-implicit": 6189, "biology": 8294, "business": 7469, "chemistry": 6615, "computer science": 6245, "connections": 468, "cta": 617, "date": 491, "diagnosis": 211, "dosage": 372, "economics": 7827, "engineering": 5182, "health": 7475, "history": 5520, "lab": 3163, "law": 6454, "math": 7692, "math_comp": 1367, "molecule_captioning": 2614, "molecule_design": 4306, "name_prediction": 8201, "olympiad": 427, "other": 7797, "philosophy": 7330, "physical": 3927, "physics": 6787, "psychology": 7819, "reaction_prediction": 6768, "retrosynthesis": 6024, "risk": 1353, "severity": 282, "spatial": 572, "zebra_puzzle": 381 } }, "raw": { "benchmark": { "ChemLLMBench": 32694, "Civil Service Examination": 7677, "Data Analysis": 627, "Date Arithmetic": 9365, "GRE & GMAT": 3855, "LSAT": 17145, "Language": 550, "MCTACO": 3891, "MMLU-Pro": 102896, "Math": 3263, "MedCalcBench": 10453, "MenatQA": 12674, "OmniMath": 31272, "Reasoning": 1056, "SAT": 7752, "SciBench": 6728, "TempReason": 12318, "TimeDial": 6412, "TimeQA": 13681, "TruthQuest": 20037 }, "model_id": { "DK-R1-Dist-Qwen-1.5B": 16108, "DK-R1-Dist-Qwen-14B": 16108, "DK-R1-Dist-Qwen-32B": 16069, "DK-R1-Dist-Qwen-7B": 16108, "gemini-2.5-flash": 15785, "gemini-3-flash": 16045, "gemini-3.1-pro": 15981, "gpt-35-turbo": 16100, "gpt-5.2": 15050, "gpt4o": 16106, "llama3d1-405b": 16108, "llama3d2-11b": 16108, "llama3d2-1b": 16107, "llama3d2-3b": 16108, "llama3d2-90b": 16108, "llama4-17B-128E": 16041, "o1-mini": 16108, "o1_re=low": 16108, "o3-mini": 16090 }, "num_examples": 304346, "target_binary": { "CORRECT": 201528, "INCORRECT": 102818 }, "target_score": { "1": 89963, "2": 12855, "3": 13029, "4": 12351, "5": 176148 }, "task": { "AMPS_Hard": 1292, "AQuA-RAT": 3855, "Algebra": 6331, "Applied Mathematics": 5697, "Calculus": 565, "Chemistry": 2694, "Date Arithmetic": 9365, "Discrete Mathematics": 5861, "E": 6532, "Geometry": 6215, "I": 7045, "LSAT-AR": 3535, "LSAT-LR": 8868, "LSAT-RC": 4742, "LogiQA-en": 7677, "MCTACO": 3891, "Math": 1988, "MenatQA-Counterfactual": 2440, "MenatQA-Order": 2925, "MenatQA-Scope": 7309, "Number Theory": 6035, "Physics": 2046, "Precalculus": 568, "S": 6460, "SAT-En": 3686, "SAT-Math": 4066, "TempReason-L2": 5960, "TempReason-L3": 6358, "TimeDial": 6412, "TimeQA-explicit": 7137, "TimeQA-implicit": 6544, "biology": 8483, "business": 7783, "chemistry": 6981, "computer science": 6542, "connections": 550, "cta": 627, "date": 513, "diagnosis": 266, "dosage": 380, "economics": 8104, "engineering": 5614, "health": 7793, "history": 5729, "lab": 3357, "law": 6797, "math": 8070, "math_comp": 1480, "molecule_captioning": 3040, "molecule_design": 5585, "name_prediction": 9036, "olympiad": 491, "other": 8129, "philosophy": 7617, "physical": 4029, "physics": 7160, "psychology": 8094, "reaction_prediction": 7819, "retrosynthesis": 7214, "risk": 1587, "severity": 321, "spatial": 638, "zebra_puzzle": 418 } } }, "kept_prompt_token_length": { "max": 4093, "mean": 853.0326345394483, "min": 252, "p50": 769.0, "p75": 988.0, "p90": 1330.0, "p95": 1659.0, "p99": 2596.0 }, "kept_sequence_length": { "max": 4095, "mean": 855.0326345394483, "min": 254, "p50": 771.0, "p75": 990.0, "p90": 1332.0, "p95": 1661.0, "p99": 2598.0 }, "length_filter_warnings": [ "max_response_tokens is greater than or equal to max_seq_length; examples can pass the response cap while having no room for the formatted prompt.", "Effective prompt budget is only 0 tokens after reserving the response cap; consider increasing training.max_seq_length." ], "max_disagreement": 1, "max_response_tokens": 4096, "max_seq_length": 4096, "on_sequence_overflow": "skip", "overflowed_response_token_length": { "max": 4096, "mean": 3629.6052631578946, "min": 1582, "p50": 3875.5, "p75": 3996.5, "p90": 4051.0, "p95": 4075.6, "p99": 4091.45 }, "raw_examples": 304346, "removed_by_disagreement": 16535, "removed_by_disagreement_pct": 5.432961169195587, "removed_by_response_length": 2501, "removed_by_response_length_pct": 0.8689730413361546, "sequence_overflow_count": 152, "sequence_overflow_pct": 0.05327538466930706, "sequence_overflow_reason": "Full chat-formatted sequence exceeded max_seq_length after response filtering. This includes system prompt, question, reference answer, model response, and chat-template tokens, plus target score tokens." }, "inference_config.yaml": { "data": { "columns": { "benchmark": "benchmark", "example_id": "instance_id", "judge_1_score": "score_gpt4o", "judge_2_score": "score_sonnet", "model_id": "model_id", "question": "question", "reference_answer": "ground_truth", "response": "response", "source": "source", "task": "task" }, "filters": { "max_disagreement": 1, "max_response_tokens": 4096, "on_sequence_overflow": "skip" }, "path": "data/processed/response_scores.parquet", "prepared_dir": null, "preprocessing_num_workers": 5, "token_length_batch_size": 2048, "tokenizers_parallelism": true }, "distributed": { "backend": "nccl", "deepspeed": { "config_overrides": {}, "gradient_clipping": "auto", "offload_optimizer_device": "none", "offload_param_device": "none", "stage3_gather_16bit_weights_on_model_save": true, "zero_stage": 2 }, "enabled": true, "find_unused_parameters": false, "fsdp": { "activation_checkpointing": true, "sharding_strategy": "full_shard", "transformer_layer_cls_to_wrap": null, "use_orig_params": true }, "gradient_checkpointing": true, "mixed_precision": "bf16", "strategy": "ddp" }, "evaluation": { "length_buckets": [ 0, 256, 512, 1024, 2048, 3072, 4096, 1000000000 ] }, "hub": { "commit_message": "Upload ADeLe distilled judge", "create_pr": false, "local_checkpoint_dir": null, "max_shard_size": "5GB", "output_staging_dir": null, "private": false, "repo_id": null }, "inference": { "allow_base_model": true, "allowed_scores": [ "1", "2", "3", "4", "5" ], "batch_size": 64, "binary_threshold": 3, "generation_fallback": false, "method": "restricted_continuation_logprobs_fast", "require_adapter": false }, "model": { "adapter_path": "runs/qwen3_14b_restricted_score_ce/adapter", "attn_implementation": "sdpa", "model_name_or_path": "Qwen/Qwen3-14B", "revision": null, "thinking_mode": { "apply_if_supported": true, "enabled": false }, "trust_remote_code": true }, "project": { "output_dir": "runs/qwen3_14b_restricted_score_ce", "run_name": "qwen3_14b_restricted_score_ce", "seed": 42 }, "prompt": { "system_prompt": "You are an answer quality evaluator. You will receive a question, a reference\nanswer, and a model response, each delimited by ### headers.\n\nRate the model response on the following scale:\n1 - Surely incorrect: the response is wrong or completely misses the point.\n2 - Likely incorrect: the response has significant errors or critical omissions.\n3 - Minimally correct / sufficient: the response is partially correct or borderline.\n4 - Likely correct: the response is mostly correct with minor issues.\n5 - Surely correct: the response is fully correct and complete.\n\nRules:\n- Base your rating strictly on correctness relative to the reference answer.\n- Ignore writing style, verbosity, or formatting unless they affect correctness.\n- If the final answer is correct but the reasoning contains substantive errors, lower the rating according to the severity of those errors.\n- Output only a single integer from 1 to 5. No explanation. No punctuation.\n" }, "split": { "held_out_model": null, "lomo_validation_fraction": 0.05, "lomo_validation_max_examples": 30000, "lomo_validation_seed": 42, "mode": "fixed_by_model", "train_models": "auto_except_val_test", "validation_models": [ "gemini-3-flash", "DK-R1-Dist-Qwen-14B", "llama3d2-3b" ] }, "training": { "cache_tokenized_datasets": true, "class_weighting": null, "dtype": "bfloat16", "eval_steps": 500, "eval_subset_seed": 42, "eval_subset_size": null, "eval_subset_strategy": "stratified", "eval_subset_stratify_columns": [ "model_id", "target_score" ], "gradient_accumulation_steps": 4, "learning_rate": 3e-05, "length_column_name": "length", "load_in_4bit": true, "logging_steps": 10, "lora_alpha": 64, "lora_dropout": 0.0, "lora_r": 32, "loss": { "class_weights": null, "lambda_binary": 0.5, "type": "ce_5way" }, "lr_scheduler_type": "cosine", "max_grad_norm": 1.0, "max_seq_length": 4096, "num_train_epochs": 1, "objective": "restricted_score_ce", "optim": "adamw_8bit", "packing": false, "per_device_eval_batch_size": 2, "per_device_train_batch_size": 2, "resume_from_checkpoint": null, "save_steps": 500, "save_total_limit": 10, "score_class_weights": null, "seed": 42, "target_modules": "auto", "train_sampling_strategy": "random", "warmup_ratio": 0.03, "weight_decay": 0.0 } }, "length_statistics.json": { "num_examples": 285158, "prompt_token_length": { "max": 4093, "mean": 853.0326345394483, "min": 252, "p50": 769.0, "p75": 988.0, "p90": 1330.0, "p95": 1659.0, "p99": 2596.0 }, "response_token_length": { "max": 3803, "mean": 397.55135047938336, "min": 0, "p50": 307.0, "p75": 508.0, "p90": 789.0, "p95": 1116.0, "p99": 2162.0 }, "sequence_length": { "max": 4095, "mean": 855.0326345394483, "min": 254, "p50": 771.0, "p75": 990.0, "p90": 1332.0, "p95": 1661.0, "p99": 2598.0 }, "target_token_length": { "max": 1, "mean": 1.0, "min": 1, "p50": 1.0, "p75": 1.0, "p90": 1.0, "p95": 1.0, "p99": 1.0 } }, "run_metadata.json": { "distributed": { "backend": "nccl", "deepspeed": { "config_overrides": {}, "gradient_clipping": "auto", "offload_optimizer_device": "none", "offload_param_device": "none", "stage3_gather_16bit_weights_on_model_save": true, "zero_stage": 2 }, "effective_global_batch_size": 64, "enabled": true, "find_unused_parameters": false, "fsdp": { "activation_checkpointing": true, "sharding_strategy": "full_shard", "transformer_layer_cls_to_wrap": null, "use_orig_params": true }, "gradient_accumulation_steps": 4, "gradient_checkpointing": true, "launcher": "torchrun", "mixed_precision": "bf16", "per_device_train_batch_size": 2, "strategy": "ddp", "world_size": 8 }, "effective_global_batch_size": 64, "evaluation_enabled": true, "git_commit": "70ddf4477faac49d4c1d9eb87cbefb58fab7e4a0", "package_versions": { "accelerate": "1.13.0", "bitsandbytes": "0.49.2", "datasets": "4.3.0", "deepspeed": "unavailable", "peft": "0.19.1", "python": "3.11.15", "torch": "2.7.1+cu118", "transformers": "5.5.0", "trl": "0.24.0", "unsloth": "2026.4.8" }, "score_class_weights": null, "score_token_ids": [ 16, 17, 18, 19, 20 ], "source_split_counts": { "test": 0, "train": 239420, "validation": 45738 }, "training_backend": "transformers_peft", "training_examples": 239420, "training_loss": { "class_weights": null, "lambda_binary": 0.5, "type": "ce_5way" }, "training_mode": "standard", "training_objective": "restricted_score_ce", "validation_full_examples": 45738, "validation_monitor_examples": 45738, "warmup": { "total_optimization_steps": 3741, "warmup_ratio": 0.03, "warmup_steps": 113 }, "world_size": 8 }, "score_tokenization_report.json": [ { "num_tokens": 1, "score": "1", "token_ids": [ 16 ], "tokens": [ "1" ] }, { "num_tokens": 1, "score": "2", "token_ids": [ 17 ], "tokens": [ "2" ] }, { "num_tokens": 1, "score": "3", "token_ids": [ 18 ], "tokens": [ "3" ] }, { "num_tokens": 1, "score": "4", "token_ids": [ 19 ], "tokens": [ "4" ] }, { "num_tokens": 1, "score": "5", "token_ids": [ 20 ], "tokens": [ "5" ] } ], "split_report.json": { "test": { "examples": 0, "models": [], "num_models": 0 }, "train": { "examples": 239420, "models": [ "DK-R1-Dist-Qwen-1.5B", "DK-R1-Dist-Qwen-32B", "DK-R1-Dist-Qwen-7B", "gemini-2.5-flash", "gemini-3.1-pro", "gpt-35-turbo", "gpt-5.2", "gpt4o", "llama3d1-405b", "llama3d2-11b", "llama3d2-1b", "llama3d2-90b", "llama4-17B-128E", "o1-mini", "o1_re=low", "o3-mini" ], "num_models": 16 }, "validation": { "examples": 45738, "models": [ "DK-R1-Dist-Qwen-14B", "gemini-3-flash", "llama3d2-3b" ], "num_models": 3 } }, "train_metrics.json": { "epoch": 1.0, "total_flos": 2.185247058507386e+19, "train_loss": 0.6326745578038823, "train_runtime": 128852.3134, "train_samples_per_second": 1.858, "train_steps_per_second": 0.029 }, "validation_trainer_metrics.json": { "epoch": 1.0, "eval_binary_accuracy": 0.9893523984433076, "eval_binary_macro_f1": 0.9880204659010172, "eval_confidence_mean": 0.9603744832085106, "eval_confidence_p50": 0.9991409778594971, "eval_confidence_p90": 0.9999487400054932, "eval_expected_calibration_error_10bin": 0.004097698586483999, "eval_f1_correct": 0.9920149535162078, "eval_f1_incorrect": 0.9840259782858267, "eval_f1_score_1": 0.9774369762155193, "eval_f1_score_2": 0.38461538461538464, "eval_f1_score_3": 0.5938009787928223, "eval_f1_score_4": 0.730398069963812, "eval_f1_score_5": 0.989033961060818, "eval_false_negative_rate_correct": 0.009138552243694727, "eval_false_positive_rate_correct": 0.013677012098895318, "eval_loss": 0.09880143404006958, "eval_num_examples": 45738, "eval_ordinal_accuracy": 0.963924963924964, "eval_ordinal_macro_f1": 0.7350570741296714, "eval_ordinal_mae": 0.04779395688486598, "eval_precision_correct": 0.9931711481007256, "eval_precision_incorrect": 0.98173964264677, "eval_precision_score_1": 0.9729145558932792, "eval_precision_score_2": 0.41139240506329117, "eval_precision_score_3": 0.6159052453468697, "eval_precision_score_4": 0.7802835051546392, "eval_precision_score_5": 0.9858030795310072, "eval_pred_binary_counts": { "CORRECT": 30459, "INCORRECT": 15279 }, "eval_pred_score_counts": { "1": 14805, "2": 474, "3": 591, "4": 1552, "5": 28316 }, "eval_recall_correct": 0.9908614477563052, "eval_recall_incorrect": 0.9863229879011047, "eval_recall_score_1": 0.9820016362148896, "eval_recall_score_2": 0.3611111111111111, "eval_recall_score_3": 0.573228346456693, "eval_recall_score_4": 0.6865079365079365, "eval_recall_score_5": 0.9922860900785611, "eval_runtime": 4341.2041, "eval_samples_per_second": 10.536, "eval_score_entropy_mean": 0.10761874169111252, "eval_score_margin_mean": 6.721137046813965, "eval_steps_per_second": 0.659, "eval_support_correct": 30530, "eval_support_incorrect": 15208, "eval_support_score_1": 14668, "eval_support_score_2": 540, "eval_support_score_3": 635, "eval_support_score_4": 1764, "eval_support_score_5": 28131, "eval_target_score_counts": { "1": 14668, "2": 540, "3": 635, "4": 1764, "5": 28131 }, "eval_within_1_accuracy": 0.993244129607766 } }, "base_model": "Qwen/Qwen3-14B", "binary_threshold": 3, "git_commit": "8a6e08f6f071ce154a27e2718a55ad33895348c6", "max_seq_length": 4096, "package_versions": { "accelerate": "1.13.0", "bitsandbytes": "0.49.2", "datasets": "4.3.0", "deepspeed": "unavailable", "peft": "0.19.1", "python": "3.11.15", "torch": "2.7.1+cu118", "transformers": "5.5.0", "trl": "0.24.0", "unsloth": "2026.4.8" }, "repo_id": "adgomant/adele-judge-qwen3-14-cre", "run_dir": "runs/qwen3_14b_restricted_score_ce", "thinking_mode": { "apply_if_supported": true, "enabled": false }, "training_objective": "restricted_score_ce" }