| method: lora_sdpo |
| base_model_id: mistralai/Mistral-7B-Instruct-v0.3 |
| seed: 42 |
| exp_name: tooluse_baseline |
| git_commit: 8b979a30de6dfbf3b5a1052e42d8c0453b214d3f |
| dataset: lasgroup/SDPO |
| dataset_slug: sdpo_tooluse |
| manifest_path: /capstor/scratch/cscs/tkwiecinski/amr-fma/train/Mistral-7B-Instruct-v0.3/lora_sdpo/sdpo_tooluse/tooluse_baseline__s42/manifest.yaml |
| tags: |
| phase: P1 |
| domain: tool_use |
| hyperparams: |
| model: |
| base_model_id: mistralai/Mistral-7B-Instruct-v0.3 |
| model_family: mistral |
| target_modules: |
| - q_proj |
| - k_proj |
| - v_proj |
| - o_proj |
| - gate_proj |
| - up_proj |
| - down_proj |
| dataset: |
| name: lasgroup/SDPO |
| split: train |
| text_field: prompt |
| max_samples: 50 |
| eval_samples: 256 |
| config: null |
| domain: tool_use |
| slug: sdpo_tooluse |
| format: tooluse |
| min_level: null |
| sequence: |
| max_length: 2048 |
| packing: true |
| lora: |
| r: 16 |
| alpha: 32 |
| dropout: 0.05 |
| target_modules: |
| - q_proj |
| - k_proj |
| - v_proj |
| - o_proj |
| - gate_proj |
| - up_proj |
| - down_proj |
| optimization: |
| num_train_epochs: 1 |
| per_device_batch_size: 1 |
| gradient_accumulation_steps: 2 |
| learning_rate: 5.0e-05 |
| warmup_ratio: 0.0 |
| weight_decay: 0.0 |
| lr_scheduler_type: cosine |
| max_grad_norm: 1.0 |
| checkpointing: |
| num_checkpoints: 8 |
| save_total_limit: 64 |
| schedule: log |
| save_steps: null |
| runtime: |
| logging_steps: 20 |
| bf16: true |
| gradient_checkpointing: true |
| wandb: true |
| wandb_project: amr-fma-train |
| hf_push: true |
| hf_org: tkwiecinski |
| hf_visibility: public |
| force_restart: false |
| sdpo: |
| reward: tooluse_match |
| num_generations: 4 |
| generation_batch_size: 32 |
| steps_per_generation: null |
| max_prompt_length: 2048 |
| max_completion_length: 1024 |
| max_reprompt_len: 10240 |
| distillation_alpha: 1.0 |
| distillation_topk: null |
| distillation_weight: 1.0 |
| distillation_is_clip: 2.0 |
| full_logit_distillation: false |
| policy_loss_mode: distillation_only |
| teacher_regularization: ema |
| teacher_update_rate: null |
| success_reward_threshold: 1.0 |
| use_successful_as_teacher: true |
| include_environment_feedback: false |
| feedback_column: null |
| beta: 0.0 |
| epsilon: 0.2 |
| scale_rewards: group |
| mask_truncated_completions: true |
| dump_rollouts: true |
| temperature: 1.3 |
| evaluation: |
| enabled: false |
| eval_steps: null |
| strategy: null |
| prompt_style: |
| system_prompt: null |
| style: none |
| final_adapter_path: /capstor/scratch/cscs/tkwiecinski/amr-fma/train/Mistral-7B-Instruct-v0.3/lora_sdpo/sdpo_tooluse/tooluse_baseline__s42/adapter_final |
| total_steps: 80 |
| checkpoints: |
| - step: 1 |
| dir: /capstor/scratch/cscs/tkwiecinski/amr-fma/train/Mistral-7B-Instruct-v0.3/lora_sdpo/sdpo_tooluse/tooluse_baseline__s42/checkpoint-1 |
| artifact: /capstor/scratch/cscs/tkwiecinski/amr-fma/train/Mistral-7B-Instruct-v0.3/lora_sdpo/sdpo_tooluse/tooluse_baseline__s42/checkpoint-1 |
| metadata: |
| source: trainer_on_save |
| metrics: |
| eval_loss: -0.110709 |
| eval_runtime: 357.3819 |
| eval_samples_per_second: 0.028 |
| eval_steps_per_second: 0.008 |
| eval_perplexity: 0.895199 |
| hf_revision: step-00001 |
| hf_commit: e2f3cdfd0dc7c5eaef1b6e32eeba8ce0f2738a13 |
| - step: 3 |
| dir: /capstor/scratch/cscs/tkwiecinski/amr-fma/train/Mistral-7B-Instruct-v0.3/lora_sdpo/sdpo_tooluse/tooluse_baseline__s42/checkpoint-3 |
| artifact: /capstor/scratch/cscs/tkwiecinski/amr-fma/train/Mistral-7B-Instruct-v0.3/lora_sdpo/sdpo_tooluse/tooluse_baseline__s42/checkpoint-3 |
| metadata: |
| source: trainer_on_save |
| metrics: |
| eval_loss: -0.182218 |
| eval_runtime: 129.1822 |
| eval_samples_per_second: 0.077 |
| eval_steps_per_second: 0.023 |
| eval_perplexity: 0.83342 |
| hf_revision: step-00003 |
| hf_commit: 14ada0c42e7468a240dd9c465f092df6049c9f1f |
| - step: 6 |
| dir: /capstor/scratch/cscs/tkwiecinski/amr-fma/train/Mistral-7B-Instruct-v0.3/lora_sdpo/sdpo_tooluse/tooluse_baseline__s42/checkpoint-6 |
| artifact: /capstor/scratch/cscs/tkwiecinski/amr-fma/train/Mistral-7B-Instruct-v0.3/lora_sdpo/sdpo_tooluse/tooluse_baseline__s42/checkpoint-6 |
| metadata: |
| source: trainer_on_save |
| metrics: |
| eval_loss: -0.201897 |
| eval_runtime: 168.0925 |
| eval_samples_per_second: 0.059 |
| eval_steps_per_second: 0.018 |
| eval_perplexity: 0.817179 |
| hf_revision: step-00006 |
| hf_commit: e1de10c2b970b009c1c4ff5a24e73295a881e8e1 |
| - step: 12 |
| dir: /capstor/scratch/cscs/tkwiecinski/amr-fma/train/Mistral-7B-Instruct-v0.3/lora_sdpo/sdpo_tooluse/tooluse_baseline__s42/checkpoint-12 |
| artifact: /capstor/scratch/cscs/tkwiecinski/amr-fma/train/Mistral-7B-Instruct-v0.3/lora_sdpo/sdpo_tooluse/tooluse_baseline__s42/checkpoint-12 |
| metadata: |
| source: trainer_on_save |
| metrics: |
| eval_loss: -0.213826 |
| eval_runtime: 194.3596 |
| eval_samples_per_second: 0.051 |
| eval_steps_per_second: 0.015 |
| eval_perplexity: 0.807489 |
| hf_revision: step-00012 |
| hf_commit: aa4049dcc138e67c77db5635e3e48d95f8659e2c |
| - step: 22 |
| dir: /capstor/scratch/cscs/tkwiecinski/amr-fma/train/Mistral-7B-Instruct-v0.3/lora_sdpo/sdpo_tooluse/tooluse_baseline__s42/checkpoint-22 |
| artifact: /capstor/scratch/cscs/tkwiecinski/amr-fma/train/Mistral-7B-Instruct-v0.3/lora_sdpo/sdpo_tooluse/tooluse_baseline__s42/checkpoint-22 |
| metadata: |
| source: trainer_on_save |
| metrics: |
| eval_loss: -0.185596 |
| eval_runtime: 160.5654 |
| eval_samples_per_second: 0.062 |
| eval_steps_per_second: 0.019 |
| eval_perplexity: 0.830609 |
| hf_revision: step-00022 |
| hf_commit: a7d4b112d07005296fcb168cd62429d8895b417d |
| - step: 42 |
| dir: /capstor/scratch/cscs/tkwiecinski/amr-fma/train/Mistral-7B-Instruct-v0.3/lora_sdpo/sdpo_tooluse/tooluse_baseline__s42/checkpoint-42 |
| artifact: /capstor/scratch/cscs/tkwiecinski/amr-fma/train/Mistral-7B-Instruct-v0.3/lora_sdpo/sdpo_tooluse/tooluse_baseline__s42/checkpoint-42 |
| metadata: |
| source: trainer_on_save |
| metrics: |
| eval_loss: -0.143644 |
| eval_runtime: 126.8726 |
| eval_samples_per_second: 0.079 |
| eval_steps_per_second: 0.024 |
| eval_perplexity: 0.866196 |
| hf_revision: step-00042 |
| hf_commit: 3b69c61dd752549a688bd3314bccb5a0d61bf43e |
| - step: 79 |
| dir: /capstor/scratch/cscs/tkwiecinski/amr-fma/train/Mistral-7B-Instruct-v0.3/lora_sdpo/sdpo_tooluse/tooluse_baseline__s42/checkpoint-79 |
| artifact: /capstor/scratch/cscs/tkwiecinski/amr-fma/train/Mistral-7B-Instruct-v0.3/lora_sdpo/sdpo_tooluse/tooluse_baseline__s42/checkpoint-79 |
| metadata: |
| source: trainer_on_save |
| metrics: |
| eval_loss: -0.102835 |
| eval_runtime: 180.9068 |
| eval_samples_per_second: 0.055 |
| eval_steps_per_second: 0.017 |
| eval_perplexity: 0.902275 |
| hf_revision: step-00079 |
| hf_commit: d00134a32d6216850e2e8998d26f23c1aadcee32 |
| - step: 80 |
| dir: /capstor/scratch/cscs/tkwiecinski/amr-fma/train/Mistral-7B-Instruct-v0.3/lora_sdpo/sdpo_tooluse/tooluse_baseline__s42/checkpoint-80 |
| artifact: /capstor/scratch/cscs/tkwiecinski/amr-fma/train/Mistral-7B-Instruct-v0.3/lora_sdpo/sdpo_tooluse/tooluse_baseline__s42/checkpoint-80 |
| metadata: |
| source: trainer_on_save |
| metrics: |
| eval_loss: -0.098016 |
| eval_runtime: 143.2853 |
| eval_samples_per_second: 0.07 |
| eval_steps_per_second: 0.021 |
| eval_perplexity: 0.906635 |
| hf_revision: step-00080 |
| hf_commit: 40d40918461f82a7ac828cf0438ecd437e626de5 |
| wandb_run_id: 9pmpx1of |
| wandb_eval_run_ids: {} |
| hf_repo_id: tkwiecinski/amr-fma-Mistral-7B-Instruct-v0.3-lora_sdpo-sdpo_tooluse-tooluse_baseline-s42 |
|
|
