Instructions to use prodigyhuh/atomicvision-hard-recall-micro-boost-lora with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use prodigyhuh/atomicvision-hard-recall-micro-boost-lora with PEFT:
from peft import PeftModel from transformers import AutoModelForCausalLM base_model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen3-1.7B") model = PeftModel.from_pretrained(base_model, "prodigyhuh/atomicvision-hard-recall-micro-boost-lora") - Notebooks
- Google Colab
- Kaggle
| { | |
| "base_model": "Qwen/Qwen3-1.7B", | |
| "adapter": "/tmp/atomicvision_publish_runner/output/train/checkpoint-1", | |
| "episodes_per_difficulty": 32, | |
| "seed_start": 10000, | |
| "seed_policy": { | |
| "sft_train": { | |
| "start": 1000, | |
| "stop": 4000 | |
| }, | |
| "grpo_train": { | |
| "start": 4000, | |
| "stop": 8000 | |
| }, | |
| "heldout_eval": { | |
| "start": 10000, | |
| "stop": 11000 | |
| } | |
| }, | |
| "heldout_seed_enforced": true, | |
| "max_tool_steps": 3, | |
| "max_new_tokens": 180, | |
| "modes": [ | |
| "strict" | |
| ], | |
| "results": { | |
| "medium": { | |
| "baseline_prior_submit": { | |
| "episodes": 32, | |
| "mean_reward": 4.65651721875, | |
| "mean_f1": 0.80580365625, | |
| "mean_mae": 0.0244615625, | |
| "mean_steps": 2.0, | |
| "mean_scan_cost": 1.5, | |
| "done_rate": 1.0, | |
| "tool_failure_rate": 0.0, | |
| "mean_repeated_tool_calls": 0.0, | |
| "strict_tool_call_pass_rate": 1.0, | |
| "normalized_tool_call_pass_rate": 1.0, | |
| "normalized_tool_call_repair_rate": 0.0, | |
| "first_action_valid_rate": 1.0, | |
| "first_action_ask_prior_rate": 1.0, | |
| "submit_action_rate": 1.0, | |
| "mean_identity_reward": 3.2232142812499998, | |
| "mean_concentration_reward": 2.23155896875, | |
| "mean_confidence_reward": 0.270494, | |
| "mean_false_positive_penalty": -0.1875, | |
| "mean_missed_defect_penalty": -0.28125, | |
| "mean_timeout_penalty": 0.0, | |
| "mean_outcome_reward_total": 5.72526725, | |
| "mean_penalty_total": -1.06875 | |
| }, | |
| "strict_adapter": { | |
| "episodes": 32, | |
| "mean_reward": 4.50648265625, | |
| "mean_f1": 0.789137, | |
| "mean_mae": 0.027124218749999998, | |
| "mean_steps": 2.0, | |
| "mean_scan_cost": 1.5, | |
| "done_rate": 1.0, | |
| "tool_failure_rate": 0.0, | |
| "mean_repeated_tool_calls": 0.0, | |
| "strict_tool_call_pass_rate": 1.0, | |
| "normalized_tool_call_pass_rate": 1.0, | |
| "normalized_tool_call_repair_rate": 0.0, | |
| "first_action_valid_rate": 1.0, | |
| "first_action_ask_prior_rate": 1.0, | |
| "submit_action_rate": 1.0, | |
| "mean_identity_reward": 3.156547625, | |
| "mean_concentration_reward": 2.16269634375, | |
| "mean_confidence_reward": 0.29348875, | |
| "mean_false_positive_penalty": -0.1875, | |
| "mean_missed_defect_penalty": -0.31875, | |
| "mean_timeout_penalty": 0.0, | |
| "mean_outcome_reward_total": 5.61273271875, | |
| "mean_penalty_total": -1.10625 | |
| }, | |
| "strict_failures": [] | |
| }, | |
| "hard": { | |
| "baseline_prior_submit": { | |
| "episodes": 32, | |
| "mean_reward": 5.01651990625, | |
| "mean_f1": 0.85153328125, | |
| "mean_mae": 0.02220903125, | |
| "mean_steps": 2.0, | |
| "mean_scan_cost": 1.5, | |
| "done_rate": 1.0, | |
| "tool_failure_rate": 0.0, | |
| "mean_repeated_tool_calls": 0.0, | |
| "strict_tool_call_pass_rate": 1.0, | |
| "normalized_tool_call_pass_rate": 1.0, | |
| "normalized_tool_call_repair_rate": 0.0, | |
| "first_action_valid_rate": 1.0, | |
| "first_action_ask_prior_rate": 1.0, | |
| "submit_action_rate": 1.0, | |
| "mean_identity_reward": 3.40613278125, | |
| "mean_concentration_reward": 2.3444645625, | |
| "mean_confidence_reward": 0.53779759375, | |
| "mean_false_positive_penalty": -0.109375, | |
| "mean_missed_defect_penalty": -0.5625, | |
| "mean_timeout_penalty": 0.0, | |
| "mean_outcome_reward_total": 6.2883949375, | |
| "mean_penalty_total": -1.2718749999999999 | |
| }, | |
| "strict_adapter": { | |
| "episodes": 32, | |
| "mean_reward": 4.714775875, | |
| "mean_f1": 0.8206800937500001, | |
| "mean_mae": 0.02552296875, | |
| "mean_steps": 2.0, | |
| "mean_scan_cost": 1.5, | |
| "done_rate": 1.0, | |
| "tool_failure_rate": 0.0, | |
| "mean_repeated_tool_calls": 0.0, | |
| "strict_tool_call_pass_rate": 1.0, | |
| "normalized_tool_call_pass_rate": 1.0, | |
| "normalized_tool_call_repair_rate": 0.0, | |
| "first_action_valid_rate": 1.0, | |
| "first_action_ask_prior_rate": 1.0, | |
| "submit_action_rate": 1.0, | |
| "mean_identity_reward": 3.282720125, | |
| "mean_concentration_reward": 2.243760375, | |
| "mean_confidence_reward": 0.5257955, | |
| "mean_false_positive_penalty": -0.09375, | |
| "mean_missed_defect_penalty": -0.64375, | |
| "mean_timeout_penalty": 0.0, | |
| "mean_outcome_reward_total": 6.052276, | |
| "mean_penalty_total": -1.3375 | |
| }, | |
| "strict_failures": [] | |
| } | |
| } | |
| } |