Instructions to use a3ilab-llm-uncertainty/gptoss_20b_all_zhtw_lr5e-7_ep1_16_64_128 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use a3ilab-llm-uncertainty/gptoss_20b_all_zhtw_lr5e-7_ep1_16_64_128 with PEFT:

from peft import PeftModel
from transformers import AutoModelForCausalLM

base_model = AutoModelForCausalLM.from_pretrained("openai/gpt-oss-20b")
model = PeftModel.from_pretrained(base_model, "a3ilab-llm-uncertainty/gptoss_20b_all_zhtw_lr5e-7_ep1_16_64_128")

Transformers

How to use a3ilab-llm-uncertainty/gptoss_20b_all_zhtw_lr5e-7_ep1_16_64_128 with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="a3ilab-llm-uncertainty/gptoss_20b_all_zhtw_lr5e-7_ep1_16_64_128")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("a3ilab-llm-uncertainty/gptoss_20b_all_zhtw_lr5e-7_ep1_16_64_128", dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use a3ilab-llm-uncertainty/gptoss_20b_all_zhtw_lr5e-7_ep1_16_64_128 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "a3ilab-llm-uncertainty/gptoss_20b_all_zhtw_lr5e-7_ep1_16_64_128"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "a3ilab-llm-uncertainty/gptoss_20b_all_zhtw_lr5e-7_ep1_16_64_128",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/a3ilab-llm-uncertainty/gptoss_20b_all_zhtw_lr5e-7_ep1_16_64_128

SGLang

How to use a3ilab-llm-uncertainty/gptoss_20b_all_zhtw_lr5e-7_ep1_16_64_128 with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "a3ilab-llm-uncertainty/gptoss_20b_all_zhtw_lr5e-7_ep1_16_64_128" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "a3ilab-llm-uncertainty/gptoss_20b_all_zhtw_lr5e-7_ep1_16_64_128",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "a3ilab-llm-uncertainty/gptoss_20b_all_zhtw_lr5e-7_ep1_16_64_128" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "a3ilab-llm-uncertainty/gptoss_20b_all_zhtw_lr5e-7_ep1_16_64_128",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use a3ilab-llm-uncertainty/gptoss_20b_all_zhtw_lr5e-7_ep1_16_64_128 with Docker Model Runner:
```
docker model run hf.co/a3ilab-llm-uncertainty/gptoss_20b_all_zhtw_lr5e-7_ep1_16_64_128
```

gptoss_20b_all_zhtw_lr5e-7_ep1_16_64_128

This model is a fine-tuned version of openai/gpt-oss-20b on the multi_turn_miss_func_zh_tw_function_mix_sharegpt_clean1, the multi_turn_miss_param_zh_tw_function_mix_sharegpt_clean1, the multi_turn_zh_tw_function_mix_sharegpt, the irrelevance_zh_tw_sharegpt3000 and the apigen_zhtwV3_remove_sys_sharegpt datasets. It achieves the following results on the evaluation set:

Loss: 1.2229

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: 5e-07
train_batch_size: 4
eval_batch_size: 4
seed: 42
gradient_accumulation_steps: 32
total_train_batch_size: 128
optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED 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: 15
num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss
1.3029	0.0673	10	1.3584
1.25	0.1347	20	1.3548
1.3156	0.2020	30	1.3417
1.2243	0.2694	40	1.3255
1.1586	0.3367	50	1.3090
1.1214	0.4040	60	1.2897
1.1924	0.4714	70	1.2696
1.1589	0.5387	80	1.2572
1.0763	0.6061	90	1.2471
1.1637	0.6734	100	1.2360
1.123	0.7407	110	1.2302
1.091	0.8081	120	1.2263
1.1514	0.8754	130	1.2232
1.129	0.9428	140	1.2212

Framework versions

PEFT 0.18.1
Transformers 4.57.6
Pytorch 2.8.0+cu128
Datasets 4.0.0
Tokenizers 0.22.2

Downloads last month: 1

Model tree for a3ilab-llm-uncertainty/gptoss_20b_all_zhtw_lr5e-7_ep1_16_64_128

Base model

openai/gpt-oss-20b

Adapter

(227)

this model