RichardErkhov/hpcgroup_-_hpc-coder-v2-6.7b-gguf

Quantization made by Richard Erkhov.

Github

Discord

Request more models

hpc-coder-v2-6.7b - GGUF

• Model creator: https://huggingface.co/hpcgroup/
• Original model: https://huggingface.co/hpcgroup/hpc-coder-v2-6.7b/

Name	Quant method	Size
hpc-coder-v2-6.7b.Q2_K.gguf	Q2_K	2.36GB
hpc-coder-v2-6.7b.IQ3_XS.gguf	IQ3_XS	2.61GB
hpc-coder-v2-6.7b.IQ3_S.gguf	IQ3_S	2.75GB
hpc-coder-v2-6.7b.Q3_K_S.gguf	Q3_K_S	2.75GB
hpc-coder-v2-6.7b.IQ3_M.gguf	IQ3_M	2.9GB
hpc-coder-v2-6.7b.Q3_K.gguf	Q3_K	3.07GB
hpc-coder-v2-6.7b.Q3_K_M.gguf	Q3_K_M	3.07GB
hpc-coder-v2-6.7b.Q3_K_L.gguf	Q3_K_L	3.35GB
hpc-coder-v2-6.7b.IQ4_XS.gguf	IQ4_XS	3.4GB
hpc-coder-v2-6.7b.Q4_0.gguf	Q4_0	3.56GB
hpc-coder-v2-6.7b.IQ4_NL.gguf	IQ4_NL	3.59GB
hpc-coder-v2-6.7b.Q4_K_S.gguf	Q4_K_S	3.59GB
hpc-coder-v2-6.7b.Q4_K.gguf	Q4_K	3.8GB
hpc-coder-v2-6.7b.Q4_K_M.gguf	Q4_K_M	3.8GB
hpc-coder-v2-6.7b.Q4_1.gguf	Q4_1	3.95GB
hpc-coder-v2-6.7b.Q5_0.gguf	Q5_0	4.33GB
hpc-coder-v2-6.7b.Q5_K_S.gguf	Q5_K_S	4.33GB
hpc-coder-v2-6.7b.Q5_K.gguf	Q5_K	4.46GB
hpc-coder-v2-6.7b.Q5_K_M.gguf	Q5_K_M	4.46GB
hpc-coder-v2-6.7b.Q5_1.gguf	Q5_1	4.72GB
hpc-coder-v2-6.7b.Q6_K.gguf	Q6_K	5.15GB
hpc-coder-v2-6.7b.Q8_0.gguf	Q8_0	6.67GB

Original model description:

library_name: transformers tags: - code - hpc - parallel - axonn datasets: - hpcgroup/hpc-instruct - ise-uiuc/Magicoder-OSS-Instruct-75K - nickrosh/Evol-Instruct-Code-80k-v1 language: - en pipeline_tag: text-generation

HPC-Coder-v2

The HPC-Coder-v2-6.7b model is an HPC code LLM fine-tuned on an instruction dataset catered to common HPC topics such as parallelism, optimization, accelerator porting, etc. This version is a fine-tuning of the Deepseek Coder 6.7b model. It is fine-tuned on the hpc-instruct, oss-instruct, and evol-instruct datasets. We utilized the distributed training library AxoNN to fine-tune in parallel across many GPUs.

HPC-Coder-v2-1.3b and HPC-Coder-v2-6.7b are two of the most capable open-source LLMs for parallel and HPC code generation. HPC-Coder-v2-6.7b is the best performing LLM under 30b parameters on the ParEval parallel code generation benchmark in terms of correctness and performance. It scores similarly to 34B and commercial models like Phind-V2 and GPT-4 on parallel code generation.

Using HPC-Coder-v2

The model is provided as a standard huggingface model with safetensor weights. It can be used with transformers pipelines, vllm, or any other standard model inference framework. HPC-Coder-v2 is an instruct model and prompts need to be formatted as instructions for best results. It was trained with the following instruct template:

Below is an instruction that describes a task. Write a response that appropriately completes the request.

### Instruction:
{instruction}

### Response:

Quantized Models

4 and 8 bit quantized weights are available in the GGUF format for use with llama.cpp. The 4 bit model requires ~3.8 GB memory and can be found here. The 8 bit model requires ~7.1 GB memory and can be found here. Further information on how to use them with llama.cpp can be found in its documentation.