What are the hardware requirements for Deepseek Coder 6.7B Instruct 3.0bpw H6 EXL2 2?

Deepseek Coder 6.7B Instruct 3.0bpw H6 EXL2 2 requires approximately 2.8 GB of VRAM and supports a context window of 16K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Deepseek Coder 6.7B Instruct 3.0bpw H6 EXL2 2 and how large is it?

Deepseek Coder 6.7B Instruct 3.0bpw H6 EXL2 2 is developed by LoneStriker, a model with 7b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

Where can I download or evaluate Deepseek Coder 6.7B Instruct 3.0bpw H6 EXL2 2?

Deepseek Coder 6.7B Instruct 3.0bpw H6 EXL2 2 is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

Deepseek Coder 6.7B Instruct 3.0bpw H6 EXL2 2 by LoneStriker — VRAM 2.8GB, 16K context

Name: Deepseek Coder 6.7B Instruct 3.0bpw H6 EXL2 2
Author: LoneStriker

Deepseek Coder 6.7B Instruct 3.0bpw H6 EXL2 2 is an open-source language model by LoneStriker. Features: 7b LLM, VRAM: 2.8GB, Context: 16K, License: other, Quantized, Instruction-Based, Code Generating, LLM Explorer Score: 0.1.

Codegen Conversational Endpoints compatible Exl2 Instruct Llama Pytorch Quantized Region:us Safetensors

Model Card on HF 🤗: https://huggingface.co/LoneStriker/deepseek-coder-6.7b-instruct-3.0bpw-h6-exl2-2

Deepseek Coder 6.7B Instruct 3.0bpw H6 EXL2 2 Parameters and Internals

Model Type

code generation, text generation

Supported Languages

languages_supported (/languages_supported/), proficiency_levels (/proficiency_levels/)

Training Details

Data Sources:

project-level code corpus

Data Volume:

2T tokens

Methodology:

fill-in-the-blank task, using a window size of 16K

Context Length:

16000

LLM Name	Deepseek Coder 6.7B Instruct 3.0bpw H6 EXL2 2
Repository 🤗	https://huggingface.co/LoneStriker/deepseek-coder-6.7b-instruct-3.0bpw-h6-exl2-2
Model Size	7b
Required VRAM	2.8 GB
Updated	2026-08-04
Maintainer	LoneStriker
Model Type	llama
Instruction-Based	Yes
Model Files	2.8 GB
Quantization Type	exl2
Generates Code	Yes
Model Architecture	LlamaForCausalLM
License	other
Context Length	16384
Model Max Length	16384
Transformers Version	4.34.1
Tokenizer Class	LlamaTokenizerFast
Beginning of Sentence Token	<｜begin▁of▁sentence｜>
End of Sentence Token	<\|EOT\|>
Vocabulary Size	32256
Torch Data Type	bfloat16

Rank the Deepseek Coder 6.7B Instruct 3.0bpw H6 EXL2 2 capabilities

Have you tried this model? Rate its performance — this feedback helps the ML community find the right model for their needs.

Instruction Following and Task Automation
Factuality and Completeness of Knowledge
Censorship and Alignment
Data Analysis and Insight Generation
Text Generation
Text Summarization and Feature Extraction
Code Generation
Multi-Language Support and Translation

Best Alternatives to Deepseek Coder 6.7B Instruct 3.0bpw H6 EXL2 2

Best Alternatives	Context / RAM	Downloads	Likes
...ruct Solidity Bnb 4bit Smashed	16K / 4.2 GB	6	0
...B Instruct Hf Bnb 4bit Smashed	16K / 4.2 GB	6	0
CodelLama7B Inst DPO 7K Mlx	16K / 4.2 GB	49	4
...eLlama 7B Instruct Hf 4bit MLX	16K / 4.2 GB	84	3
...6.7B Instruct 8.0bpw H8 EXL2 2	16K / 6.8 GB	7	2
... 7B Instruct Nf4 Fp16 Upscaled	16K / 13.5 GB	9	0
CodeLlama 7B Instruct Fp16	16K / 13.5 GB	24	7
...Llama 7B Instruct Bf16 Sharded	16K / 13.5 GB	29	1
...B Instruct V1.5 6.0bpw H6 EXL2	4K / 5.7 GB	8	2
...B Instruct V1.5 8.0bpw H8 EXL2	4K / 7.3 GB	8	1

Note: green Score (e.g. "73.2") means that the model is better than LoneStriker/deepseek-coder-6.7b-instruct-3.0bpw-h6-exl2-2.