| method: lora_sft |
| base_model_id: meta-llama/Llama-3.1-8B-Instruct |
| seed: 43 |
| exp_name: p1_sft_math_tooluse |
| git_commit: 8b979a30de6dfbf3b5a1052e42d8c0453b214d3f |
| dataset: lasgroup/SDPO |
| dataset_slug: sdpo_tooluse |
| manifest_path: /capstor/scratch/cscs/tkwiecinski/amr-fma/train/Llama-3.1-8B-Instruct/lora_sft/sdpo_tooluse/p1_sft_math_tooluse__s43/manifest.yaml |
| tags: |
| phase: P1 |
| domain: tool_use |
| hyperparams: |
| model: |
| base_model_id: meta-llama/Llama-3.1-8B-Instruct |
| model_family: llama3 |
| 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: null |
| 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: 3 |
| per_device_batch_size: 4 |
| gradient_accumulation_steps: 8 |
| learning_rate: 0.0002 |
| warmup_ratio: 0.05 |
| weight_decay: 0.01 |
| 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: null |
| evaluation: |
| enabled: true |
| eval_steps: 200 |
| strategy: steps |
| prompt_style: null |
| final_adapter_path: /capstor/scratch/cscs/tkwiecinski/amr-fma/train/Llama-3.1-8B-Instruct/lora_sft/sdpo_tooluse/p1_sft_math_tooluse__s43/adapter_final |
| total_steps: 114 |
| checkpoints: |
| - step: 1 |
| dir: /capstor/scratch/cscs/tkwiecinski/amr-fma/train/Llama-3.1-8B-Instruct/lora_sft/sdpo_tooluse/p1_sft_math_tooluse__s43/checkpoint-1 |
| artifact: /capstor/scratch/cscs/tkwiecinski/amr-fma/train/Llama-3.1-8B-Instruct/lora_sft/sdpo_tooluse/p1_sft_math_tooluse__s43/checkpoint-1 |
| metadata: |
| source: trainer_on_save |
| metrics: |
| eval_loss: 1.272672 |
| eval_runtime: 21.1731 |
| eval_samples_per_second: 3.826 |
| eval_steps_per_second: 0.992 |
| eval_perplexity: 3.570381 |
| hf_revision: step-00001 |
| hf_commit: 0d4986498ea3fc44e19079d35eee157277715f70 |
| - step: 3 |
| dir: /capstor/scratch/cscs/tkwiecinski/amr-fma/train/Llama-3.1-8B-Instruct/lora_sft/sdpo_tooluse/p1_sft_math_tooluse__s43/checkpoint-3 |
| artifact: /capstor/scratch/cscs/tkwiecinski/amr-fma/train/Llama-3.1-8B-Instruct/lora_sft/sdpo_tooluse/p1_sft_math_tooluse__s43/checkpoint-3 |
| metadata: |
| source: trainer_on_save |
| metrics: |
| eval_loss: 1.206122 |
| eval_runtime: 11.0618 |
| eval_samples_per_second: 7.323 |
| eval_steps_per_second: 1.898 |
| eval_perplexity: 3.340504 |
| hf_revision: step-00003 |
| hf_commit: 5848980eb5187a5b6a3c445cb411af3bd64e9680 |
| - step: 7 |
| dir: /capstor/scratch/cscs/tkwiecinski/amr-fma/train/Llama-3.1-8B-Instruct/lora_sft/sdpo_tooluse/p1_sft_math_tooluse__s43/checkpoint-7 |
| artifact: /capstor/scratch/cscs/tkwiecinski/amr-fma/train/Llama-3.1-8B-Instruct/lora_sft/sdpo_tooluse/p1_sft_math_tooluse__s43/checkpoint-7 |
| metadata: |
| source: trainer_on_save |
| metrics: |
| eval_loss: 0.781388 |
| eval_runtime: 11.0652 |
| eval_samples_per_second: 7.32 |
| eval_steps_per_second: 1.898 |
| eval_perplexity: 2.184502 |
| hf_revision: step-00007 |
| hf_commit: 3312a9ced0a76447d646da154b3009d3b37778b5 |
| - step: 14 |
| dir: /capstor/scratch/cscs/tkwiecinski/amr-fma/train/Llama-3.1-8B-Instruct/lora_sft/sdpo_tooluse/p1_sft_math_tooluse__s43/checkpoint-14 |
| artifact: /capstor/scratch/cscs/tkwiecinski/amr-fma/train/Llama-3.1-8B-Instruct/lora_sft/sdpo_tooluse/p1_sft_math_tooluse__s43/checkpoint-14 |
| metadata: |
| source: trainer_on_save |
| metrics: |
| eval_loss: 0.466285 |
| eval_runtime: 11.0629 |
| eval_samples_per_second: 7.322 |
| eval_steps_per_second: 1.898 |
| eval_perplexity: 1.594061 |
| hf_revision: step-00014 |
| hf_commit: 53df359874697e9811bf984311da87e0bf65903b |
| - step: 29 |
| dir: /capstor/scratch/cscs/tkwiecinski/amr-fma/train/Llama-3.1-8B-Instruct/lora_sft/sdpo_tooluse/p1_sft_math_tooluse__s43/checkpoint-29 |
| artifact: /capstor/scratch/cscs/tkwiecinski/amr-fma/train/Llama-3.1-8B-Instruct/lora_sft/sdpo_tooluse/p1_sft_math_tooluse__s43/checkpoint-29 |
| metadata: |
| source: trainer_on_save |
| metrics: |
| eval_loss: 0.379769 |
| eval_runtime: 11.0591 |
| eval_samples_per_second: 7.324 |
| eval_steps_per_second: 1.899 |
| eval_perplexity: 1.461946 |
| hf_revision: step-00029 |
| hf_commit: 4531078bb439378d5a6bfdfdc88e4d4eef143a6d |
| - step: 57 |
| dir: /capstor/scratch/cscs/tkwiecinski/amr-fma/train/Llama-3.1-8B-Instruct/lora_sft/sdpo_tooluse/p1_sft_math_tooluse__s43/checkpoint-57 |
| artifact: /capstor/scratch/cscs/tkwiecinski/amr-fma/train/Llama-3.1-8B-Instruct/lora_sft/sdpo_tooluse/p1_sft_math_tooluse__s43/checkpoint-57 |
| metadata: |
| source: trainer_on_save |
| metrics: |
| eval_loss: 0.271018 |
| eval_runtime: 11.0913 |
| eval_samples_per_second: 7.303 |
| eval_steps_per_second: 1.893 |
| eval_perplexity: 1.311298 |
| hf_revision: step-00057 |
| hf_commit: cb4e9fbb49d9ae4a48b123f49378459279fdc0fd |
| - step: 114 |
| dir: /capstor/scratch/cscs/tkwiecinski/amr-fma/train/Llama-3.1-8B-Instruct/lora_sft/sdpo_tooluse/p1_sft_math_tooluse__s43/checkpoint-114 |
| artifact: /capstor/scratch/cscs/tkwiecinski/amr-fma/train/Llama-3.1-8B-Instruct/lora_sft/sdpo_tooluse/p1_sft_math_tooluse__s43/checkpoint-114 |
| metadata: |
| source: trainer_on_save |
| metrics: |
| eval_loss: 0.148939 |
| eval_runtime: 11.059 |
| eval_samples_per_second: 7.324 |
| eval_steps_per_second: 1.899 |
| eval_perplexity: 1.160602 |
| hf_revision: step-00114 |
| hf_commit: f05f56d523c19166bda3280bb90fc989a0c28f23 |
| wandb_run_id: 4l3ivdio |
| wandb_eval_run_ids: {} |
| hf_repo_id: tkwiecinski/amr-fma-Llama-3.1-8B-Instruct-lora_sft-sdpo_tooluse-p1_sft_math_tooluse-s43 |
|
|
