chat: fix case where template accepts type content only by ngxson · Pull Request #19419 · ggml-org/llama.cpp

ngxson · 2026-02-07T17:20:12Z

Fix chat template of PaddleOCR-VL, which requires content to be an array (see #18825)

This should be able to handle these case:

Template supports ONLY string content
Template supports ONLY array content
Template supports both string AND array content

ngxson · 2026-02-09T15:41:34Z

Pinging @pwilkin @aldehir @CISC if you can have a quick look on this (currently a roadblock for PaddleOCR). Thanks!

pwilkin · 2026-02-09T17:59:19Z

+                    }
+                });
+            }
+            printf("%s\n", jmsg.dump().c_str());


Remove or convert to LOG_*

pwilkin · 2026-02-09T18:03:14Z

 // Parses a JSON array of messages in OpenAI's chat completion API format.
 std::vector<common_chat_msg> common_chat_msgs_parse_oaicompat(const nlohmann::ordered_json & messages);
+
+// DEPRECATED: only used in tests


Wouldn't it be a good idea just to replace common_chat_msgs_to_json_oaicompat with the body of the new render_message_to_json (with the extended signature with full caps) and make the old common_chat_msgs_to_json_oaicompat call it with just the single cap set?

Good idea, I implemented in 6edb330 , test-chat is OK

* chat: fix case where template accepts type content only * rm stray log * reuse render_message_to_json

@tugot17

… and new jinja template engine (ggml-org#1369) --------- Co-authored-by: Piotr Wilkin <piotr.wilkin@syndatis.com> common : add nemotron 3 parsing (ggml-org#18077) common : add parser for ministral/mistral large 3/devstral 2 (ggml-org#17713) common : default content to an empty string (ggml-org#18485) chat: make tool description and parameters optional per OpenAI spec (ggml-org#18478) Per the OpenAI API specification, both 'description' and 'parameters' fields in tool function definitions are optional. Previously, the parser would throw an exception if these fields were missing. Attempts to fix ggml-org#17667 common : implement new jinja template engine (ggml-org#18462) --------- Co-authored-by: Alde Rojas <hello@alde.dev> Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> jinja: correct member access rule (ggml-org#18905) jinja : fix lexing of float literals with sign (ggml-org#18901) jinja : add missing tojson filter for bool (ggml-org#18900) jinja : attribute support for join, map and sort (ggml-org#18883) jinja : fix object item order (and properly implement dictsort) (ggml-org#18904) tests : add test-jinja -py option for cross-checking (ggml-org#18906) Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com> --------- Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com> ci : run test-jinja -py on high perf [no ci] (ggml-org#18916) jinja : fix undefined keys and attributes and int/float as bool (ggml-org#18924) jinja: support none|string (ggml-org#18995) Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com> Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com> --------- Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com> jinja : implement mixed type object keys (ggml-org#18955) --------- Co-authored-by: Xuan Son Nguyen <son@huggingface.co> jinja : undefined should be treated as sequence/iterable (return string/array) by filters/tests (ggml-org#19147) `tojson` is not a supported `undefined` filter keep it DRY and fix some types jinja : do not pass empty tools and add some none filters (ggml-org#19176) jinja : add unordered_map include to value.h [no ci] (ggml-org#19205) jinja : add missing 'in' test to template engine (ggml-org#19004) (ggml-org#19239) The jinja template parser was missing the 'in' test from global_builtins(), causing templates using reject("in", ...), select("in", ...), or 'x is in(y)' to fail with "selectattr: unknown test 'in'". This broke tool-calling for Qwen3-Coder and any other model whose chat template uses the 'in' test. Added test_is_in supporting array, string, and object containment checks, mirroring the existing 'in' operator logic in runtime.cpp. Includes test cases for all three containment types plus reject/select filter usage. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> --------- Co-authored-by: Sid Mohan <sidmohan0@users.noreply.github.com> Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com> Co-authored-by: Xuan Son Nguyen <son@huggingface.co> Add Jinja support for "indent" string filter (ggml-org#19529) Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com> Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com> --------- Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com> add vendor refactor chat server : support preserving reasoning_content in assistant message (ggml-org#18994) chat : fix translategemma crash on common_chat_format_example (ggml-org#19019) chat: fix language input for translategemma (ggml-org#19052) Co-authored-by: Aldehir Rojas <hello@alde.dev> --------- Co-authored-by: Aldehir Rojas <hello@alde.dev> chat: fix case where template accepts type content only (ggml-org#19419) mtmd : chat : Fix extra \n between text and media marker (ggml-org#19595) Thanks to @tugot17 for detecting and reporting the issue. For vision models (e.g. LFM2.5-VL-1.6B and Qwen/Qwen3-VL-4B-Instruct) `llama-mtmd-cli` produces identical output to HF implementation. However `llama-server` doesn't. I traced it down to extra newline inserted after `<__media__>`. This happens in `to_json_oaicompat`, that treats media markers as text and joins all parts with `\n` separator. PR introduces new type `media_marker` and uses it for media markers. Extra logic is added to prevent insertion of newlines before and after media markers. With this change number of input tokens is identical to HF implementation and as a result the output is also identical. I explored other ways to address the issue * remove completely `\n` between text parts in `to_json_oaicompat` * merge text messages in server-common.cpp before sending them to `to_json_oaicompat` Please propose alternative ways of fixing this issue. Co-authored-by: Piotr Wilkin (ilintar) <piotr.wilkin@syndatis.com> --------- Co-authored-by: Piotr Wilkin (ilintar) <piotr.wilkin@syndatis.com> common : merge qwen3-coder and nemotron nano 3 parsers (ggml-org#19765) common : fix improper trimming in XML parser on complete message (ggml-org#19805) Co-authored-by: Jules LEIDELINGER <11395311+julio75012@users.noreply.github.com> jinja: correct stats for tojson and string filters (ggml-org#19785) jinja : correct default size for string slices (ggml-org#19913) common : handle unicode during partial json parsing (ggml-org#16526) common : fix json schema with '\' in literals (ggml-org#17307) add back qwen_coder_xml and mirothinker Co-authored-by: Aldehir Rojas <hello@alde.dev>

chat: fix case where template accepts type content only

7c0c73d

ngxson requested review from CISC, aldehir and pwilkin as code owners February 7, 2026 17:20

ngxson mentioned this pull request Feb 7, 2026

model: Add PaddleOCR-VL model support #18825

Merged

github-actions Bot added the jinja parser Issues related to the jinja parser label Feb 7, 2026

ownia mentioned this pull request Feb 9, 2026

common : convert string contents to arrays if template requires typed content #19156

Closed

pwilkin approved these changes Feb 9, 2026

View reviewed changes

ngxson added 2 commits February 9, 2026 19:42

rm stray log

0de61b5

reuse render_message_to_json

6edb330

CISC approved these changes Feb 9, 2026

View reviewed changes

aldehir approved these changes Feb 9, 2026

View reviewed changes

ngxson merged commit 98e57ca into ggml-org:master Feb 9, 2026
73 of 75 checks passed

liparetejas pushed a commit to liparetejas/llama.cpp that referenced this pull request Feb 23, 2026

chat: fix case where template accepts type content only (ggml-org#19419)

1304166

* chat: fix case where template accepts type content only * rm stray log * reuse render_message_to_json

bartowski1182 pushed a commit to bartowski1182/llama.cpp that referenced this pull request Mar 2, 2026

chat: fix case where template accepts type content only (ggml-org#19419)

943f6f0

* chat: fix case where template accepts type content only * rm stray log * reuse render_message_to_json

LETS-BEE mentioned this pull request Mar 9, 2026

Eval bug: Jinja Template Parsing Error in TranslateGemma #20305

Open

Seunghhon pushed a commit to Seunghhon/llama.cpp that referenced this pull request Apr 26, 2026

chat: fix case where template accepts type content only (ggml-org#19419)

fc71976

* chat: fix case where template accepts type content only * rm stray log * reuse render_message_to_json

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

chat: fix case where template accepts type content only#19419

chat: fix case where template accepts type content only#19419
ngxson merged 3 commits intoggml-org:masterfrom
ngxson:xsn/chat_fix_typed_content_only

ngxson commented Feb 7, 2026 •

edited

Loading

Uh oh!

ngxson commented Feb 9, 2026

Uh oh!

pwilkin Feb 9, 2026

Uh oh!

pwilkin Feb 9, 2026

Uh oh!

ngxson Feb 9, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

Conversation

ngxson commented Feb 7, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

ngxson commented Feb 9, 2026

Uh oh!

pwilkin Feb 9, 2026

Choose a reason for hiding this comment

Uh oh!

pwilkin Feb 9, 2026

Choose a reason for hiding this comment

Uh oh!

ngxson Feb 9, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

ngxson commented Feb 7, 2026 •

edited

Loading