Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Models: Add Qwen2-1.5B-Instruct #2759

Merged
merged 2 commits into from
Jul 29, 2024
Merged

Models: Add Qwen2-1.5B-Instruct #2759

merged 2 commits into from
Jul 29, 2024

Conversation

ThiloteE
Copy link
Collaborator

@ThiloteE ThiloteE commented Jul 27, 2024

Adds a models3.json entry for Qwen2-1.5B-Instruct

Description of Model

It is a tiny bilingual model and at the date of writing with very strong results in benchmarks (for its parameter size). It supports a context of up to 32768. Because of its model size it has very fast responses, even when doing inference on CPU. This LLM is LITERALLY for all. Since the model fits into 4GB of RAM (just barely, if the Operating System and other apps also need RAM) or alternatively into 3GB of VRAM, this will be the workhorse of the desperate and hardware poor.

  • The model was trained/finetuned on English and Chinese language
  • License: Apache 2.0

Personal Impression:

I got the impression the model is very task focused and this is the reason, why I chose Below is an instruction that describes a task. Write a response that appropriately completes the request. as system prompt. Since the model is relatively small, its responses may seem not very coherent or intelligent, but it works surprisingly well with GPT4All's LocalDocs feature. It is like the model was made for RAG. Its long context adds to that. It mainly will appeal to English and Chinese speaking users.

Checklist before requesting a review

  • I have performed a self-review of my code.
  • If it is a core feature, I have added thorough tests.
  • I have added thorough documentation for my code.
  • I have tagged PR with relevant project labels. I acknowledge that a PR without labels may be dismissed.
  • If this PR addresses a bug, I have provided both a screenshot/video of the original bug and the working solution.

@ThiloteE
Copy link
Collaborator Author

I have added this model at the location "order": "z",, because I fear there might be merge conflicts with #2750

@ThiloteE ThiloteE added models models.json This requires a change to the official model list. labels Jul 27, 2024
@ThiloteE ThiloteE marked this pull request as ready for review July 27, 2024 19:53
@ThiloteE
Copy link
Collaborator Author

VAGOsolutions confirm its RAG capabilities in their German RAG benchmark:
image

@cosmic-snow
Copy link
Collaborator

I've downloaded it and checked some fields, and they're all fine: md5sum, name, filename, filesize, quant, type, parameters

I have not looked at their site/blog to verify the templates, however a quick test with them went well.

@manyoso manyoso merged commit e45685b into main Jul 29, 2024
6 of 12 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
models.json This requires a change to the official model list. models
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants