[Stacked] Remove existing StoredInference PyO3 type #4594
+5,113
−7,836
Conversation
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
890f2dc to
ab29765
Compare
900ea55 to
3f095b0
Compare
ab29765 to
3efd6b8
Compare
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
Comment on lines
386
to
+390
| } else { | ||
| deserialize_from_pyobj(py, obj) | ||
| let wire: StoredInference = deserialize_from_pyobj(py, obj)?; | ||
| let storage = match wire.to_storage(config) { | ||
| Ok(s) => s, | ||
| Err(e) => return Err(tensorzero_core_error(py, &e.to_string())?), |
There was a problem hiding this comment.
Preserve datapoint deserialization in render_samples
The new deserialize_from_stored_sample branch now treats every non‑Datapoint object as a StoredInference by deserializing with StoredInference and calling to_storage. Previously, arbitrary Python objects matching the StoredSample schema (including manually constructed datapoints) were deserialized as a StoredSampleItem. After this change, passing a dataclass or dict for a stored datapoint to experimental_render_samples will fail with Failed to deserialize JSON to StoredInference instead of being accepted. This breaks the documented ability to render datapoints that were created in Python without going through the Rust bindings.
Useful? React with 👍 / 👎.
4c5e1e4 to
cd26846
Compare
3ceea3b to
e4349f9
Compare
3efd6b8 to
70bba41
Compare
e4349f9 to
c8955bf
Compare
* add limit and offset support * pass values as parameters * explicit reverse chronological ordering * refine docstrings * update test
* Improve error information in rate limiting Fix #3556 * Fix * Fix * Fix
* Fix * Fix
…ools (#4599) * wip * implemented migration 0041 to add tool columns * updated migration to fix CH issue and add prtovider_tool colummn * updated migration for ModelInference and dynamic_provider_tools * ModelInference -> BatchModelInference * fixed documentation; * renamed Tool -> ClientSideFunctionTool * refactored tool.rs (except tests) to new write pattern * fixed tests in tool file * mostly refactored function.rs * everything compiles * added tool deserializers * almost all tests fixed * tool_config -> tool_params * unit tests pass * temporarily removed rollback instrucitons * built bindings * re-enabled the rollback; added a sleep * for real re-enable the rollback * try a longer sleep * removed sleep * fixed migration 0041 by modifying EpisodeByID views to use corrected syntax * removed unnecessary function in eval helper tests * removed stray comment * cleaned up stray comments * wip * all dataset tests pass * I think all e2e tests pass * clippies * removed stray comment * fixed some outstanding clickhouse tests * fixed remaining clickhouse tests * fixed issue with legacy endpoint * Regenerate ModelInferenceCache fixtures * removed stray comment * removed unused serializer * fixed one more stray query * fixed issues with tests that I created * wip * added additional documentation of tool related types * made provider_tools mandatory across codebase * added a comment for why this is a custom deserialier * cleaned up test helper * fixed remaining PR issues * fixed test for union all * fixed all unit tests * built bindings * removed new_for_test and fixed UI type issue * Regenerate ModelInferenceCache fixtures * added pyo3 getters to ClientSideFunctionTool * fixed tests that broke with PR comments * wip * fixed Python tests * fixed Python tests * fixed a couple more failing client tests * fixed batch tests by adding one more deserializer case * added some more unit tests * fixed bad test assertion * fixed bad test assertion * batch tests should pass now * wip on mocking * wip on mocking * tests run but fail * wip on macros * full coverage but definitely broken * wip * some more of these passed * some more of these passed * fixed a couple more * fixed a couple more * fixed a couple more * fixed a couple more * all mock tests pass * clippies pass * removed stray config.toml alias * reverted changes to openai.rs * reverted changes to test checking functions * removed unnecessary toml file * updated CI settings to use mocked tests in merge queue and live ones daily * removed extra field from model, fixed clippies * fixed clippies * set up config for provider types for mock batch URL * wip * cleaned up batch URL mock injection for gcp * injected batch API base to OpenAI config construction * removed all unneccessary loose assertions for mock batch tests * fixed all tests for batch + mock * removed stray &s * cfged out a step * skip extra fields for bindings * fixed lockfile * skip loading fixtures in mock batch tests * updated internal tool handling to allow sending all with a limited list of allowed tools * changed GCP vertex preparation of tools to named function * updated GCP vertex provider to send all but then specify allowed tools * updated google AI studio to send but limit tools * updated groq to send all tools with allowed * implemented all allowed tools in prod * updated all providers * added unit tests for anthropic and strict tools avaialble * unit tests pass * fixed responses api * AllAllowedTools -> OnlyAllowedTools * AllowedToolsChoice::Explicit * fireworks works * stuff works besides vllm / sgl * added handling for sglang and vllm * fixed issue with python bindings * all tests passing * fixed clippies * removed extra schemas * removed nextest changes * removed nextest changes * cleaned up imports & tests * cleaned up clippies; * removed stray change * cleaned up diff some more * fixed generated types and typechecked * fixed PR comments * fixed more PR comments --------- Co-authored-by: TensorZero Bot <github-actions[bot]@users.noreply.github.com> Co-authored-by: Claude <[email protected]>
* Use fresh ClickHouse db for test_count_datasets This prevents concurrent test executions from affecting the final count * Add 'clickhouse' to the test name to give it a longer timeout
* Clean up docs/tutorial * Clean up docs/tutorial
tests Pyright
70bba41 to
eadc19c
Compare
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.
No description provided.