Instructions to use xverse/XVERSE-13B-Chat-GPTQ-Int8 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use xverse/XVERSE-13B-Chat-GPTQ-Int8 with Transformers:

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

pipe = pipeline("text-generation", model="xverse/XVERSE-13B-Chat-GPTQ-Int8", trust_remote_code=True)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("xverse/XVERSE-13B-Chat-GPTQ-Int8", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use xverse/XVERSE-13B-Chat-GPTQ-Int8 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "xverse/XVERSE-13B-Chat-GPTQ-Int8"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "xverse/XVERSE-13B-Chat-GPTQ-Int8",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/xverse/XVERSE-13B-Chat-GPTQ-Int8

SGLang

How to use xverse/XVERSE-13B-Chat-GPTQ-Int8 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 "xverse/XVERSE-13B-Chat-GPTQ-Int8" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "xverse/XVERSE-13B-Chat-GPTQ-Int8",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "xverse/XVERSE-13B-Chat-GPTQ-Int8" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "xverse/XVERSE-13B-Chat-GPTQ-Int8",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use xverse/XVERSE-13B-Chat-GPTQ-Int8 with Docker Model Runner:
```
docker model run hf.co/xverse/XVERSE-13B-Chat-GPTQ-Int8
```

XVERSE-13B-Chat-GPTQ-Int8

File size: 1,240 Bytes

532b134

{
  "_name_or_path": "/mnt/cfs1/zhangxu/xverse_ckpts/xland/xverse2-13b-fanyule847491-1204_xland-1226/pretrain_MP1/iter_0000846/mp_rank_00",
  "architectures": [
    "XverseForCausalLM"
  ],
  "auto_map": {
    "AutoConfig": "configuration_xverse.XverseConfig",
    "AutoModelForCausalLM": "modeling_xverse.XverseForCausalLM"
  },
  "bos_token_id": 2,
  "eos_token_id": 3,
  "hidden_act": "silu",
  "hidden_size": 5120,
  "initializer_range": 0.02,
  "intermediate_size": 13824,
  "max_position_embeddings": 8192,
  "max_tokenizer_truncation": 6144,
  "model_type": "xverse",
  "num_attention_heads": 40,
  "num_hidden_layers": 40,
  "num_key_value_heads": 40,
  "pad_token_id": 1,
  "quantization_config": {
    "bits": 8,
    "damp_percent": 0.01,
    "desc_act": true,
    "group_size": 128,
    "is_marlin_format": false,
    "model_file_base_name": "gptq_model-8bit-128g",
    "model_name_or_path": "/mnt/llm_dataset/xverse_models/gptq/XVERSE-13B-Chat-GPTQ-Int8",
    "quant_method": "gptq",
    "static_groups": true,
    "sym": true,
    "true_sequential": true
  },
  "rms_norm_eps": 1e-06,
  "tie_word_embeddings": false,
  "torch_dtype": "float16",
  "transformers_version": "4.38.2",
  "use_cache": true,
  "vocab_size": 100534
}