Implement sharded model support in model_manager#1348
Closed
james-martinez wants to merge 8 commits intolemonade-sdk:mainfrom
Closed
Implement sharded model support in model_manager#1348james-martinez wants to merge 8 commits intolemonade-sdk:mainfrom
james-martinez wants to merge 8 commits intolemonade-sdk:mainfrom
Conversation
Added support for sharded model variants by automatically discovering and adding remaining parts to the download queue.
Member
|
@claude please review. In particular, does this PR have any unaddressed corner cases? |
Contributor
|
Claude Code is working… I'll analyze this and get back to you. |
Contributor
Author
I'd like to address all cases. Focus on using a particular gguf quant in any folder |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Added support for sharded model variants by automatically discovering and adding remaining parts to the download queue.
This adds support to download a particular quantization when all the gguf files are in 1 directory.
checkpoint would be NVIDIA-Nemotron-3-Super-120B-A12B:Q4_K_M
Rule 3 (Original): It checks for exactly Q4_K_M.gguf. -> Fails.
added rule : It checks if the repo contains any file matching Q4_K_M-00001-of-. -> Success! It finds NVIDIA-Nemotron-3-Super-120B-A12B-Q4_K_M-00001-of-00003.gguf and sets it as the primary file.
Rule 4 (Folder check): It completely skips this step because it already found a match in Rule 3.
The New Bottom Logic: It inspects the primary file, sees it spans 3 parts, and automatically adds 00002 and 00003 to the download queue.
example
https://huggingface.co/lmstudio-community/NVIDIA-Nemotron-3-Super-120B-A12B-GGUF/tree/main
contains
Q4_K_M Quantization
NVIDIA-Nemotron-3-Super-120B-A12B-Q4_K_M-00001-of-00003.gguf
NVIDIA-Nemotron-3-Super-120B-A12B-Q4_K_M-00002-of-00003.gguf
NVIDIA-Nemotron-3-Super-120B-A12B-Q4_K_M-00003-of-00003.gguf
Q6_K Quantization
NVIDIA-Nemotron-3-Super-120B-A12B-Q6_K-00001-of-00003.gguf
NVIDIA-Nemotron-3-Super-120B-A12B-Q6_K-00002-of-00003.gguf
NVIDIA-Nemotron-3-Super-120B-A12B-Q6_K-00003-of-00003.gguf
Q8_0 Quantization
NVIDIA-Nemotron-3-Super-120B-A12B-Q8_0-00001-of-00004.gguf
NVIDIA-Nemotron-3-Super-120B-A12B-Q8_0-00002-of-00004.gguf
NVIDIA-Nemotron-3-Super-120B-A12B-Q8_0-00003-of-00004.gguf
NVIDIA-Nemotron-3-Super-120B-A12B-Q8_0-00004-of-00004.gguf