RichardErkhov/Bin12345_-_AutoCoder_S_6.7B-gguf

Quantization made by Richard Erkhov.

AutoCoder_S_6.7B - GGUF

Model creator: https://huggingface.co/Bin12345/
Original model: https://huggingface.co/Bin12345/AutoCoder_S_6.7B/

Name	Quant method	Size
AutoCoder_S_6.7B.Q2_K.gguf	Q2_K	2.36GB
AutoCoder_S_6.7B.IQ3_XS.gguf	IQ3_XS	2.61GB
AutoCoder_S_6.7B.IQ3_S.gguf	IQ3_S	2.75GB
AutoCoder_S_6.7B.Q3_K_S.gguf	Q3_K_S	2.75GB
AutoCoder_S_6.7B.IQ3_M.gguf	IQ3_M	2.9GB
AutoCoder_S_6.7B.Q3_K.gguf	Q3_K	3.07GB
AutoCoder_S_6.7B.Q3_K_M.gguf	Q3_K_M	3.07GB
AutoCoder_S_6.7B.Q3_K_L.gguf	Q3_K_L	3.35GB
AutoCoder_S_6.7B.IQ4_XS.gguf	IQ4_XS	3.4GB
AutoCoder_S_6.7B.Q4_0.gguf	Q4_0	3.56GB
AutoCoder_S_6.7B.IQ4_NL.gguf	IQ4_NL	3.59GB
AutoCoder_S_6.7B.Q4_K_S.gguf	Q4_K_S	3.59GB
AutoCoder_S_6.7B.Q4_K.gguf	Q4_K	3.8GB
AutoCoder_S_6.7B.Q4_K_M.gguf	Q4_K_M	3.8GB
AutoCoder_S_6.7B.Q4_1.gguf	Q4_1	3.95GB
AutoCoder_S_6.7B.Q5_0.gguf	Q5_0	4.33GB
AutoCoder_S_6.7B.Q5_K_S.gguf	Q5_K_S	4.33GB
AutoCoder_S_6.7B.Q5_K.gguf	Q5_K	4.46GB
AutoCoder_S_6.7B.Q5_K_M.gguf	Q5_K_M	4.46GB
AutoCoder_S_6.7B.Q5_1.gguf	Q5_1	4.72GB
AutoCoder_S_6.7B.Q6_K.gguf	Q6_K	5.15GB
AutoCoder_S_6.7B.Q8_0.gguf	Q8_0	6.67GB

Original model description:

license: apache-2.0

We introduced a new model designed for the Code generation task. It 33B version's test accuracy on the HumanEval base dataset surpasses that of GPT-4 Turbo (April 2024). (90.9% vs 90.2%).

Additionally, compared to previous open-source models, AutoCoder offers a new feature: it can automatically install the required packages and attempt to run the code until it deems there are no issues, whenever the user wishes to execute the code.

This is the 6.7B version of AutoCoder. Its base model is deepseeker-coder.

See details on the AutoCoder GitHub.

Simple test script:

model_path = ""
tokenizer = AutoTokenizer.from_pretrained(model_path)
model = AutoModelForCausalLM.from_pretrained(model_path, 
                                             device_map="auto")

HumanEval = load_dataset("evalplus/humanevalplus")

Input = "" # input your question here
 
messages=[
    { 'role': 'user', 'content': Input}
]
inputs = tokenizer.apply_chat_template(messages, add_generation_prompt=True, 
                                        return_tensors="pt").to(model.device)

outputs = model.generate(inputs, 
                        max_new_tokens=1024, 
                        do_sample=False, 
                        temperature=0.0,
                        top_p=1.0, 
                        num_return_sequences=1, 
                        eos_token_id=tokenizer.eos_token_id)

answer = tokenizer.decode(outputs[0][len(inputs[0]):], skip_special_tokens=True)

Paper: https://arxiv.org/abs/2405.14906