| | --- |
| | library_name: transformers |
| | base_model: timarni/qwen3_dpo |
| | tags: |
| | - generated_from_trainer |
| | datasets: |
| | - timarni/MNLP_STEM_IT_HARD |
| | model-index: |
| | - name: outputs/dpo_it_hard |
| | results: [] |
| | --- |
| | |
| | <!-- This model card has been generated automatically according to the information the Trainer had access to. You |
| | should probably proofread and complete it, then remove this comment. --> |
| |
|
| | [<img src="https://raw.githubusercontent.com/axolotl-ai-cloud/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>](https://github.com/axolotl-ai-cloud/axolotl) |
| | <details><summary>See axolotl config</summary> |
| |
|
| | axolotl version: `0.9.2` |
| | ```yaml |
| | base_model: timarni/qwen3_dpo |
| | # Automatically upload checkpoint and final model to HF |
| | # hub_model_id: username/custom_model_name |
| | |
| | plugins: |
| | - axolotl.integrations.cut_cross_entropy.CutCrossEntropyPlugin |
| | strict: false |
| | |
| | chat_template: qwen3 |
| | datasets: |
| | - path: timarni/MNLP_STEM_IT_HARD |
| | type: alpaca |
| | split: train |
| | |
| | shuffle_merged_datasets: true |
| | |
| | val_set_size: 0.1 |
| | output_dir: ./outputs/dpo_it_hard |
| | dataset_prepared_path: last_run_prepared |
| | |
| | sequence_len: 4096 #2048 |
| | sample_packing: true # was true -> need to check if it actually learns on the samples or not (better understand te hyperparam and event. install axolotl to debug) |
| | eval_sample_packing: false |
| | pad_to_sequence_len: true |
| | # train_on_inputs: true # NEW |
| | # group_by_length: false NEW? |
| | |
| | # To be sure that no LORA is done |
| | adapter: null |
| | lora: false |
| | merge_lora: false |
| | |
| | wandb_project: mnlp_project |
| | wandb_entity: tim-arni |
| | wandb_watch: |
| | wandb_name: dpo_it_hard |
| | wandb_log_model: |
| | |
| | gradient_accumulation_steps: 16 # 2 |
| | micro_batch_size: 2 # 1 |
| | num_epochs: 15 |
| | optimizer: adamw_torch |
| | lr_scheduler: cosine |
| | learning_rate: 0.00001 # 0.00005 |
| | # cosine_min_lr_ratio: 0.1 |
| | |
| | warmup_ratio: 0.05 |
| | weight_decay: 0.01 |
| | |
| | bf16: auto |
| | tf32: true |
| | |
| | gradient_checkpointing: offload |
| | gradient_checkpointing_kwargs: |
| | use_reentrant: false |
| | resume_from_checkpoint: |
| | logging_steps: 1 |
| | gradient_clipping: 1.0 # or max_grad_norm? |
| | flash_attention: true |
| | |
| | evals_per_epoch: 2 |
| | saves_per_epoch: 1 |
| | save_total_limit: 20 |
| | special_tokens: |
| | |
| | ``` |
| |
|
| | </details><br> |
| |
|
| | # outputs/dpo_it_hard |
| |
|
| | This model is a fine-tuned version of [timarni/qwen3_dpo](https://huggingface.co/timarni/qwen3_dpo) on the timarni/MNLP_STEM_IT_HARD dataset. |
| | It achieves the following results on the evaluation set: |
| | - Loss: 0.1297 |
| | |
| | ## Model description |
| | |
| | More information needed |
| | |
| | ## Intended uses & limitations |
| | |
| | More information needed |
| | |
| | ## Training and evaluation data |
| | |
| | More information needed |
| | |
| | ## Training procedure |
| | |
| | ### Training hyperparameters |
| | |
| | The following hyperparameters were used during training: |
| | - learning_rate: 1e-05 |
| | - train_batch_size: 2 |
| | - eval_batch_size: 2 |
| | - seed: 42 |
| | - distributed_type: multi-GPU |
| | - num_devices: 4 |
| | - gradient_accumulation_steps: 16 |
| | - total_train_batch_size: 128 |
| | - total_eval_batch_size: 8 |
| | - optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments |
| | - lr_scheduler_type: cosine |
| | - lr_scheduler_warmup_steps: 2 |
| | - num_epochs: 15.0 |
| |
|
| | ### Training results |
| |
|
| | | Training Loss | Epoch | Step | Validation Loss | |
| | |:-------------:|:------:|:----:|:---------------:| |
| | | 0.7556 | 0.3404 | 1 | 0.7317 | |
| | | 0.7451 | 0.6809 | 2 | 0.5623 | |
| | | 0.5054 | 1.0 | 3 | 0.2737 | |
| | | 0.1901 | 1.3404 | 4 | 0.1879 | |
| | | 0.1304 | 1.6809 | 5 | 0.1532 | |
| | | 0.1146 | 2.0 | 6 | 0.1421 | |
| | | 0.1046 | 2.3404 | 7 | 0.1377 | |
| | | 0.1001 | 2.6809 | 8 | 0.1353 | |
| | | 0.1009 | 3.0 | 9 | 0.1338 | |
| | | 0.0957 | 3.3404 | 10 | 0.1330 | |
| | | 0.0931 | 3.6809 | 11 | 0.1323 | |
| | | 0.0945 | 4.0 | 12 | 0.1316 | |
| | | 0.0914 | 4.3404 | 13 | 0.1312 | |
| | | 0.0894 | 4.6809 | 14 | 0.1307 | |
| | | 0.0912 | 5.0 | 15 | 0.1303 | |
| | | 0.0883 | 5.3404 | 16 | 0.1302 | |
| | | 0.0868 | 5.6809 | 17 | 0.1301 | |
| | | 0.0889 | 6.0 | 18 | 0.1299 | |
| | | 0.0864 | 6.3404 | 19 | 0.1299 | |
| | | 0.0856 | 6.6809 | 20 | 0.1298 | |
| | | 0.0878 | 7.0 | 21 | 0.1299 | |
| | | 0.0858 | 7.3404 | 22 | 0.1299 | |
| | | 0.085 | 7.6809 | 23 | 0.1298 | |
| | | 0.0874 | 8.0 | 24 | 0.1298 | |
| | | 0.0855 | 8.3404 | 25 | 0.1299 | |
| | | 0.0849 | 8.6809 | 26 | 0.1297 | |
| | | 0.0873 | 9.0 | 27 | 0.1298 | |
| | | 0.0854 | 9.3404 | 28 | 0.1297 | |
| | | 0.0849 | 9.6809 | 29 | 0.1297 | |
| | | 0.0873 | 10.0 | 30 | 0.1297 | |
| |
|
| |
|
| | ### Framework versions |
| |
|
| | - Transformers 4.51.3 |
| | - Pytorch 2.5.1+cu121 |
| | - Datasets 3.5.1 |
| | - Tokenizers 0.21.1 |
| |
|
