Repository navigation
chore(vendor): move tinyinference to ae1f09d (local runtime native tools) - #612
Conversation
…ols) Picks up build_local_runtime_chat_model_with_native_tools (tinyhumansai/tinyinference#154) so OpenHuman can turn native tool calls on for a local-openai endpoint that supports them. Co-authored-by: Medulla <medulla@tinyhumans.ai>
Tiny Sweeper reviewReviewed head: Verdict: Ready for maintainer review · Priority: none Diagnostics and supporting detailsFull summaryPull request purpose (title): chore(vendor): move tinyinference to ae1f09d (local runtime native tools). Tiny Sweeper reviewed this change across 5 lane(s) and found 0 active actionable finding(s). Detailed lane evidence and any incomplete work are listed below. Review snapshot
Completeness: Complete FeaturesNone identified with supported citations. TestsNo supported feature-to-test mapping was produced. Test execution is not inferred. FindingsNo active actionable findings. Before mergeNone. Agent review detailscritique
security
tests
commits
description
Evidence and run details
|
Codex Review Summary: ✅ CompletedThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
Co-authored-by: Medulla <medulla@tinyhumans.ai>
Why
Teeny's builder agent runs
glm-5.3-flashon Surplus through OpenHuman'slocal-openairuntime. The runtime always has native tool calls off, and with the JSON-in-tag text protocol Surplus turns the model's<tool_call>text into a call whose name is the whole JSON body, with arguments{}. Every builder job ends "abandoned: the builder never reported back".Probes against Surplus (glm-5.3-flash, OpenAI-compatible), run on the Teeny box:
tools, model asked to emit JSON-in-tag: notool_calls; the text stays text.tools=[shell], a normal ask: a clean native call,finish_reason=tool_calls, idcall_7df8…, nameshell, args{"command":"ls"}. Native tool calling works for this model on Surplus.toolssent plus "don't use native calls, write JSON-in-tag", non-stream: content came back verbatim,tool_callsnull.stream=true: only thinking chunks and no answer. Surplus attachesresponse_error: {format_error, content_validation: ["valid_think_chunk"], tool_calls_validation: [], extra_tags: ["only_thinking"]}. Surplus validates and rewrites output server-side, so the text protocol plus thinking is fragile there.What
Moves
vendor/tinyinferencetoae1f09d(tinyhumansai/tinyinference#154), which addsbuild_local_runtime_chat_model_with_native_tools. The change is additive.cargo check -p tinyagents-harness --all-targetspasses.Depends on tinyhumansai/tinyinference#154. After it merges, move the gitlink to the merge commit.
Test
No harness code changes. The regression test is in tinyhumansai/tinyinference#154 (
local_runtime_builder_can_opt_into_native_tools).Summary by CodeRabbit