fix: convert_hf_to_gguf - map new mistral-common valid_tokenizer_files output to avoid crash with --mistral-format#17712
Merged
CISC merged 5 commits intoggml-org:masterfrom Dec 3, 2025
Conversation
…istral-common versions.
Member
|
I think the correct fix is to try to import and use |
… fallback to old logic otherwise.
Contributor
Author
|
Yeah makes sense. Checking for What about this? also so there's another (kind of redundant) Tested on |
CISC
reviewed
Dec 3, 2025
Member
|
Thank you for taking the time to fix this, will merge once CI gets its act together. :) |
CISC
reviewed
Dec 3, 2025
Member
Mistral never provided chat templates before, but it would make sense to use |
CISC
approved these changes
Dec 3, 2025
khemchand-zetta
pushed a commit
to khemchand-zetta/llama.cpp
that referenced
this pull request
Dec 4, 2025
…l-format) (ggml-org#17712) * fix convert_hf_to_gguf.py failing with --mistral-format using later mistral-common versions. * use get_one_valid_tokenizer_file from mistral-common if available and fallback to old logic otherwise. * use file name instead of file path for get_one_valid_tokenizer_file. * fix --mistral-format tokenizer file failing for tokenizers in subdirectories. * move get_one_valid_tokenizer_file import to avoid nested try-except.
gabe-l-hart
added a commit
to gabe-l-hart/llama.cpp
that referenced
this pull request
Dec 4, 2025
* origin/master: server: strip content-length header on proxy (ggml-org#17734) server: move msg diffs tracking to HTTP thread (ggml-org#17740) examples : add missing code block end marker [no ci] (ggml-org#17756) common : skip model validation when --help is requested (ggml-org#17755) ggml-cpu : remove asserts always evaluating to false (ggml-org#17728) convert: use existing local chat_template if mistral-format model has one. (ggml-org#17749) cmake : simplify build info detection using standard variables (ggml-org#17423) ci : disable ggml-ci-x64-amd-* (ggml-org#17753) common: use native MultiByteToWideChar (ggml-org#17738) metal : use params per pipeline instance (ggml-org#17739) llama : fix sanity checks during quantization (ggml-org#17721) build : move _WIN32_WINNT definition to headers (ggml-org#17736) build: enable parallel builds in msbuild using MTT (ggml-org#17708) ggml-cpu: remove duplicate conditional check 'iid' (ggml-org#17650) Add a couple of file types to the text section (ggml-org#17670) convert : support latest mistral-common (fix conversion with --mistral-format) (ggml-org#17712) Use OpenAI-compatible `/v1/models` endpoint by default (ggml-org#17689) webui: Fix zero pasteLongTextToFileLen to disable conversion being overridden (ggml-org#17445)
0Marble
pushed a commit
to 0Marble/llama.cpp
that referenced
this pull request
Dec 18, 2025
…l-format) (ggml-org#17712) * fix convert_hf_to_gguf.py failing with --mistral-format using later mistral-common versions. * use get_one_valid_tokenizer_file from mistral-common if available and fallback to old logic otherwise. * use file name instead of file path for get_one_valid_tokenizer_file. * fix --mistral-format tokenizer file failing for tokenizers in subdirectories. * move get_one_valid_tokenizer_file import to avoid nested try-except.
Anico2
added a commit
to Anico2/llama.cpp
that referenced
this pull request
Jan 15, 2026
…l-format) (ggml-org#17712) * fix convert_hf_to_gguf.py failing with --mistral-format using later mistral-common versions. * use get_one_valid_tokenizer_file from mistral-common if available and fallback to old logic otherwise. * use file name instead of file path for get_one_valid_tokenizer_file. * fix --mistral-format tokenizer file failing for tokenizers in subdirectories. * move get_one_valid_tokenizer_file import to avoid nested try-except.
blime4
referenced
this pull request
in blime4/llama.cpp
Feb 5, 2026
…l-format) (#17712) * fix convert_hf_to_gguf.py failing with --mistral-format using later mistral-common versions. * use get_one_valid_tokenizer_file from mistral-common if available and fallback to old logic otherwise. * use file name instead of file path for get_one_valid_tokenizer_file. * fix --mistral-format tokenizer file failing for tokenizers in subdirectories. * move get_one_valid_tokenizer_file import to avoid nested try-except.
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.
Fixes #17691
mistral-commonupdated_filter_valid_tokenizer_filesto return additional data we don't need or expect, causing the conversion to crash when--mistral-formatis used.This change just maps the output back into the format originally expected.
Tested on
mistral-common==1.8.3andmistral-common==1.8.6... that said, there's a different issue with the new Ministral models, which is partially related to how
--mistral-formatworks:get_community_chat_templates()is invoked with--mistral-format, and the logic there results in the model getting the"unsloth-mistral-Devstral-Small-2507.jinja"template - not Ministral's localchat_template.jinja- that's only used forSpecialVocabmodels (notMistralVocab).To avoid this, users currently need to manually specify the chat template when running the model on e.g: llama-server with
--jinja --chat-template-file "./chat_template.jinja"This issue is the case regardless of this PR - as long as
--mistral-formatis used I think.