deepseek-ai/DeepSeek-V3

Fork 0

mirror of https://github.com/deepseek-ai/DeepSeek-V3.git synced 2026-11-05 00:47:48 +00:00

[GH-ISSUE #8] Is there a plan to release a smaller model? #6

New issue

Closed

opened 2026-10-26 21:10:12 +00:00 by bsakaguchi · 9 comments

bsakaguchi commented

2026-10-26 21:10:12 +00:00

Owner

Originally created by @lingyezhixing on GitHub (Dec 26, 2024).
Original GitHub issue: https://github.com/deepseek-ai/DeepSeek-V3/issues/8

Thank you very much for your work, the new V3 model is very powerful, but this model is too large. Is there a plan to release a smaller model around 14B like V2?

Originally created by @lingyezhixing on GitHub (Dec 26, 2024). Original GitHub issue: https://github.com/deepseek-ai/DeepSeek-V3/issues/8 Thank you very much for your work, the new V3 model is very powerful, but this model is too large. Is there a plan to release a smaller model around 14B like V2?

bsakaguchi

2026-10-26 21:10:12 +00:00

closed this issue
added the
stale
label

bsakaguchi commented

2026-10-26 21:10:14 +00:00

Author

Owner

@YangWang92 commented on GitHub (Dec 27, 2024):

I found that they have some configurations https://github.com/deepseek-ai/DeepSeek-V3/tree/main/inference/configs, which include 16B/236B/671B.

@YangWang92 commented on GitHub (Dec 27, 2024): I found that they have some configurations https://github.com/deepseek-ai/DeepSeek-V3/tree/main/inference/configs, which include 16B/236B/671B.

bsakaguchi commented

2026-10-26 21:10:14 +00:00

Author

Owner

@lebronjamesking commented on GitHub (Dec 27, 2024):

Looking forward to it. A small size version would be perfect.

@lebronjamesking commented on GitHub (Dec 27, 2024): Looking forward to it. A small size version would be perfect.

bsakaguchi commented

2026-10-26 21:10:15 +00:00

Author

Owner

@silentlustre commented on GitHub (Dec 27, 2024):

I found that they have some configurations https://github.com/deepseek-ai/DeepSeek-V3/tree/main/inference/configs, which include 16B/236B/671B.

in the paper, they trained the 16B and 236B models for ablation, these were not trained to completion.
i'm holding out hope that they'll go back and train them all the way

@silentlustre commented on GitHub (Dec 27, 2024): > I found that they have some configurations https://github.com/deepseek-ai/DeepSeek-V3/tree/main/inference/configs, which include 16B/236B/671B. in the paper, they trained the 16B and 236B models for ablation, these were not trained to completion. i'm holding out hope that they'll go back and train them all the way

bsakaguchi commented

2026-10-26 21:10:15 +00:00

Author

Owner

@KirariSpark commented on GitHub (Dec 29, 2024):

ese were not trained to completion.
i'm holding out hope that they'll go back and train them all the way

It will be great if the training is completed! I don't have much VRAM, 16b is all I can afford😂

@KirariSpark commented on GitHub (Dec 29, 2024): > ese were not trained to completion. > i'm holding out hope that they'll go back and train them all the way It will be great if the training is completed! I don't have much VRAM, 16b is all I can afford😂

bsakaguchi commented

2026-10-26 21:10:15 +00:00

Author

Owner

@pizts commented on GitHub (Dec 30, 2024):

Are there any plans to release int4 quantization?

@pizts commented on GitHub (Dec 30, 2024): Are there any plans to release int4 quantization?

bsakaguchi commented

2026-10-26 21:10:16 +00:00

Author

Owner

@workatcam commented on GitHub (Jan 1, 2025):

Are there any plans to release int4 quantization?

https://huggingface.co/inarikami/DeepSeek-V3-int4-TensorRT

@workatcam commented on GitHub (Jan 1, 2025): > Are there any plans to release int4 quantization? https://huggingface.co/inarikami/DeepSeek-V3-int4-TensorRT

bsakaguchi commented

2026-10-26 21:10:16 +00:00

Author

Owner

@XksA-me commented on GitHub (Jan 13, 2025):

Are there any plans to release int4 quantization?

https://huggingface.co/inarikami/DeepSeek-V3-int4-TensorRT

How should I use it? Do I need to download the model first and then run trtllm-build? Is there any pre-built file? Thanks for your reply.

@XksA-me commented on GitHub (Jan 13, 2025): > > Are there any plans to release int4 quantization? > > https://huggingface.co/inarikami/DeepSeek-V3-int4-TensorRT How should I use it? Do I need to download the model first and then run trtllm-build? Is there any pre-built file? Thanks for your reply.

bsakaguchi commented

2026-10-26 21:10:16 +00:00

Author

Owner

@github-actions[bot] commented on GitHub (Feb 14, 2025):

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. If you believe this issue is still relevant, please leave a comment to keep it open. Thank you for your contributions!

@github-actions[bot] commented on GitHub (Feb 14, 2025): This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. If you believe this issue is still relevant, please leave a comment to keep it open. Thank you for your contributions!

bsakaguchi commented

2026-10-26 21:10:17 +00:00

Author

Owner

@MrAppAndCrap commented on GitHub (Feb 25, 2025):

I'd like to see a mini version as well :)

@MrAppAndCrap commented on GitHub (Feb 25, 2025): I'd like to see a mini version as well :)

bsakaguchi referenced this issue

2026-10-26 21:29:48 +00:00

[GH-ISSUE #1255] Chinese vs English for DeepSeek Reasoning - Controlled Experiment with V4 Flash #808

bsakaguchi referenced this issue

2026-10-26 21:30:30 +00:00