ysr/hyperparam-rust-sft-lora

1---2license: other3library_name: peft4tags:5- trl6- sft7- generated_from_trainer8base_model: deepseek-ai/deepseek-coder-1.3b-base9datasets:10- generator11model-index:12- name: hyperparam-rust-sft-lora13  results: []14---15 16<!-- This model card has been generated automatically according to the information the Trainer had access to. You17should probably proofread and complete it, then remove this comment. -->18 19# hyperparam-rust-sft-lora20 21This model is a fine-tuned version of [deepseek-ai/deepseek-coder-1.3b-base](https://huggingface.co/deepseek-ai/deepseek-coder-1.3b-base) on the generator dataset.22It achieves the following results on the evaluation set:23- Loss: 0.424724 25## Model description26 27More information needed28 29## Intended uses & limitations30 31More information needed32 33## Training and evaluation data34 35More information needed36 37## Training procedure38 39### Training hyperparameters40 41The following hyperparameters were used during training:42- learning_rate: 0.000343- train_batch_size: 144- eval_batch_size: 145- seed: 4246- gradient_accumulation_steps: 3247- total_train_batch_size: 3248- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-0849- lr_scheduler_type: cosine50- lr_scheduler_warmup_ratio: 0.0551- lr_scheduler_warmup_steps: 2052- num_epochs: 553 54### Training results55 56| Training Loss | Epoch | Step | Validation Loss |57|:-------------:|:-----:|:----:|:---------------:|58| 0.7919        | 0.3   | 25   | 0.5285          |59| 0.4811        | 0.59  | 50   | 0.4738          |60| 0.4512        | 0.89  | 75   | 0.4567          |61| 0.4367        | 1.18  | 100  | 0.4465          |62| 0.4162        | 1.48  | 125  | 0.4399          |63| 0.4188        | 1.77  | 150  | 0.4352          |64| 0.4127        | 2.07  | 175  | 0.4318          |65| 0.3981        | 2.37  | 200  | 0.4296          |66| 0.3887        | 2.66  | 225  | 0.4281          |67| 0.3943        | 2.96  | 250  | 0.4258          |68| 0.3808        | 3.25  | 275  | 0.4263          |69| 0.3836        | 3.55  | 300  | 0.4251          |70| 0.3824        | 3.84  | 325  | 0.4247          |71| 0.3782        | 4.14  | 350  | 0.4246          |72| 0.377         | 4.43  | 375  | 0.4247          |73| 0.3725        | 4.73  | 400  | 0.4247          |74 75 76### Framework versions77 78- PEFT 0.10.079- Transformers 4.39.380- Pytorch 2.2.181- Datasets 2.18.082- Tokenizers 0.15.2