---
library_name: transformers
license: apache-2.0
base_model: openai/gpt-oss-20b
tags:
- generated_from_trainer
datasets:
- HuggingFaceH4/Multilingual-Thinking
model-index:
- name: outputs/gpt-oss-out-fft/
  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: `0.12.0.dev0`
```yaml
base_model: openai/gpt-oss-20b
use_kernels: true
model_quantization_config: Mxfp4Config
model_quantization_config_kwargs:
  dequantize: true
  block_size: 32        # default, matches the OCP spec
  strict: false         # fallback to fp16 on any odd layers
  
plugins:
  - axolotl.integrations.cut_cross_entropy.CutCrossEntropyPlugin

experimental_skip_move_to_device: true  # prevent OOM by NOT putting model to GPU before sharding

datasets:
  # - path: winglian/pirate-ultrachat-10k
  #   type: chat_template
  #   split: train
- message_property_mappings:
    content: content
    role: role
  path: HuggingFaceH4/Multilingual-Thinking
  trust_remote_code: false
  field_messages: messages
  type: chat_template
  
dataset_prepared_path: last_run_prepared
val_set_size: 0
output_dir: ./outputs/gpt-oss-out-fft/

sequence_len: 8192
sample_packing: true

# adapter: lora
# lora_r: 8
# lora_alpha: 16
# lora_dropout: 0.0
# lora_target_linear: true

wandb_project: gpt-oss-20b
wandb_name: multilingual-reasoning

gradient_accumulation_steps: 1
micro_batch_size: 2
num_epochs: 1

optimizer: adamw_torch_8bit
lr_scheduler: constant_with_warmup
learning_rate: 2e-5

bf16: true
tf32: true

flash_attention: true
attn_implementation: kernels-community/vllm-flash-attn3

gradient_checkpointing: true
activation_offloading: true

logging_steps: 1
saves_per_epoch: 1

warmup_ratio: 0.1

special_tokens:
eot_tokens:
  - "<|end|>"

fsdp_version: 2
fsdp_config:
  offload_params: false
  state_dict_type: SHARDED_STATE_DICT
  auto_wrap_policy: TRANSFORMER_BASED_WRAP
  transformer_layer_cls_to_wrap: GptOssDecoderLayer
  reshard_after_forward: true


```

</details><br>

# outputs/gpt-oss-out-fft/

This model is a fine-tuned version of [openai/gpt-oss-20b](https://huggingface.co/openai/gpt-oss-20b) on the HuggingFaceH4/Multilingual-Thinking dataset.

## 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: 2e-05
- train_batch_size: 2
- eval_batch_size: 2
- seed: 42
- distributed_type: multi-GPU
- num_devices: 8
- total_train_batch_size: 16
- total_eval_batch_size: 16
- optimizer: Use OptimizerNames.ADAMW_TORCH_8BIT with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: constant_with_warmup
- training_steps: 8

### Training results



### Framework versions

- Transformers 4.55.0
- Pytorch 2.8.0+cu128
- Datasets 4.0.0
- Tokenizers 0.21.4