duyntnet/Megrez-3B-Instruct-imatrix-GGUF

Quantizations of https://huggingface.co/Infinigence/Megrez-3B-Instruct

Note: you will need llama.cpp b4381 or later to run the model.

Inference Clients/UIs

• llama.cpp
• KoboldCPP
• ollama
• jan
• text-generation-webui
• GPT4All

---

From original readme

Megrez-3B-Instruct is a large language model trained by Infinigence AI. Megrez-3B aims to provide a fast inference, compact, and powerful edge-side intelligent solution through software-hardware co-design. Megrez-3B has the following advantages:

1. High Accuracy: Megrez-3B successfully compresses the capabilities of the previous 14 billion model into a 3 billion size, and achieves excellent performance on mainstream benchmarks.
2. High Speed: A smaller model does not necessarily bring faster speed. Megrez-3B ensures a high degree of compatibility with mainstream hardware through software-hardware co-design, leading an inference speedup up to 300% compared to previous models of the same accuracy.
3. Easy to Use: In the beginning, we had a debate about model design: should we design a unique but efficient model structure, or use a classic structure for ease of use? We chose the latter and adopt the most primitive LLaMA structure, which allows developers to deploy the model on various platforms without any modifications and minimize the complexity of future development.
4. Rich Applications: We have provided a fullstack WebSearch solution. Our model is functionally trained on web search tasks, enabling it to automatically determine the timing of search invocations and provide better summarization results. The complete deployment code is released on github.

Inference Parameters

• For chat, text generation, and other tasks that benefit from diversity, we recommend to use the inference parameter temperature=0.7.
• For mathematical and reasoning tasks, we recommend to use the inference parameter temperature=0.2 for better determinacy.

Huggingface

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

path = "Infinigence/Megrez-3B-Instruct"
device = "cuda"

tokenizer = AutoTokenizer.from_pretrained(path, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(path, torch_dtype=torch.bfloat16, device_map=device, trust_remote_code=True)

messages = [
    {"role": "user", "content": "How to make braised chicken in brown sauce?"},
]
model_inputs = tokenizer.apply_chat_template(messages, return_tensors="pt", add_generation_prompt=True).to(device)

model_outputs = model.generate(
    model_inputs,
    do_sample=True,
    max_new_tokens=1024,
    top_p=0.9,
    temperature=0.2
)

output_token_ids = [
    model_outputs[i][len(model_inputs[i]):] for i in range(len(model_inputs))
]

responses = tokenizer.batch_decode(output_token_ids, skip_special_tokens=True)[0]
print(responses)

vLLM Inference

• Installation

# Install vLLM with CUDA 12.1.
pip install vllm

• Example code

python inference/inference_vllm.py --model_path <hf_repo_path> --prompt_path prompts/prompt_demo.txt
from transformers import AutoTokenizer
from vllm import LLM, SamplingParams

model_name = "Infinigence/Megrez-3B-Instruct"
prompt = [{"role": "user", "content": "How to make braised chicken in brown sauce?"}]

tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
input_text = tokenizer.apply_chat_template(prompt, tokenize=False, add_generation_prompt=True)

llm = LLM(
    model=model_name,
    trust_remote_code=True,
    tensor_parallel_size=1
)
sampling_params = SamplingParams(top_p=0.9, temperature=0.2, max_tokens=1024, repetition_penalty=1.02)

outputs = llm.generate(prompts=input_text, sampling_params=sampling_params)

print(outputs[0].outputs[0].text)