Instructions to use ai-sage/GigaChat-20B-A3B-instruct-bf16 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use ai-sage/GigaChat-20B-A3B-instruct-bf16 with Transformers:

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

pipe = pipeline("text-generation", model="ai-sage/GigaChat-20B-A3B-instruct-bf16", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("ai-sage/GigaChat-20B-A3B-instruct-bf16", trust_remote_code=True, dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use ai-sage/GigaChat-20B-A3B-instruct-bf16 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "ai-sage/GigaChat-20B-A3B-instruct-bf16"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ai-sage/GigaChat-20B-A3B-instruct-bf16",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/ai-sage/GigaChat-20B-A3B-instruct-bf16

SGLang

How to use ai-sage/GigaChat-20B-A3B-instruct-bf16 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 "ai-sage/GigaChat-20B-A3B-instruct-bf16" \
    --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": "ai-sage/GigaChat-20B-A3B-instruct-bf16",
		"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 "ai-sage/GigaChat-20B-A3B-instruct-bf16" \
        --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": "ai-sage/GigaChat-20B-A3B-instruct-bf16",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use ai-sage/GigaChat-20B-A3B-instruct-bf16 with Docker Model Runner:
```
docker model run hf.co/ai-sage/GigaChat-20B-A3B-instruct-bf16
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

GigaChat-20B-A3B-instruct bf16

This model is part of the GigaChat family of Russian LLMs, based on ai-sage/GigaChat-20B-A3B-instruct. It supports a context length of 131,000 tokens.

More details are available in this habr article and the original instruct model card. The model was presented in GigaChat Family: Efficient Russian Language Modeling Through Mixture of Experts Architecture.

Example Usage with Transformers

pip install --upgrade transformers torch accelerate bitsandbytes

import torch
from transformers import AutoTokenizer, AutoModelForCausalLM, GenerationConfig

model_name = "ai-sage/GigaChat-20B-A3B-instruct-bf16"
tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(model_name, trust_remote_code=True, device_map="auto", torch_dtype=torch.bfloat16)
model.generation_config = GenerationConfig.from_pretrained(model_name)

messages = [
    {"role": "user", "content": "Докажи теорему о неподвижной точке"}
]
input_tensor = tokenizer.apply_chat_template(messages, add_generation_prompt=True, return_tensors="pt")
outputs = model.generate(input_tensor.to(model.device))

result = tokenizer.decode(outputs[0][input_tensor.shape[1]:], skip_special_tokens=False)
print(result)