Repository navigation
convert_qwen36: describe a qwen3_moe container as what it is - #1829
Merged
Merged
Conversation
The README the converter writes and its closing line said Qwen3.6-35B-A3B and Gated DeltaNet for every container, so a Qwen3-Coder container (#1828) shipped a README about a different model. A qwen3_moe container now gets its own README (attention layers, no DeltaNet, no shared expert, the expert format), and the closing line counts the DeltaNet layers instead of naming them.
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Follow-up to #1828.
tools/convert_qwen36.pywrote the same README into every container ("colibri container for Qwen3.6-35B-A3B ... incl. Gated DeltaNet"), and its closing line said "incl. Gated DeltaNet" too, so a Qwen3-Coder container described a different model. Found while publishing the converted Qwen3-Coder container on Hugging Face.qwen3_moecontainer now gets its own README: the layer count, no DeltaNet and no shared expert, and the expert format (bits and group size).All 48 layers stored (0 Gated DeltaNet)).Checked with the tiny
qwen3-coder-30bgeometry (int4 gs64) and the default Qwen3.6 geometry: the first gets the new README and0 Gated DeltaNet, the second the old README and6 Gated DeltaNet. Converter output only; the weights are unchanged.