fix: 5 critical bugs — TEE randomization, HTTP 402 hint, int dtype, inference None crash, gas estimation#201

Open

amathxbt wants to merge 27 commits intoOpenGradient:mainfrom

amathxbt:fix/critical-5-bugs-tee-402-dtype-inference-gas

Commits on Mar 27, 2026

test: add HTTP 402 hint coverage for completion, chat, and streaming
amathxbt
committed
fix: merge main into llm_test.py — restore retry/SSL tests, add 402 hint coverage, fix mock status_code
amathxbt
committed
fix: remove stale Union import from _conversions.py (convert_fixed_point_to_python now always returns np.float32)
amathxbt
committed
chore: sync .github/workflows/release.yml from upstream main
amathxbt
committed
chore: sync README.md from upstream main
amathxbt
committed
chore: sync WINDOWS_INSTALL.md from upstream main
amathxbt
committed
chore: sync docs/opengradient/client/llm.md from upstream main
amathxbt
committed
chore: sync docs/opengradient/index.md from upstream main
amathxbt
committed
chore: sync integrationtest/llm/test_llm_chat.py from upstream main
amathxbt
committed
chore: sync pyproject.toml from upstream main
amathxbt
committed
chore: sync src/opengradient/client/_utils.py from upstream main
amathxbt
committed
chore: sync src/opengradient/client/opg_token.py from upstream main
amathxbt
committed
chore: sync tests/opg_token_test.py from upstream main
amathxbt
committed
chore: sync stresstest/llm.py from upstream main
amathxbt
committed
chore: sync stresstest/utils.py from upstream main
amathxbt
committed
chore: sync uv.lock from upstream main
amathxbt
committed
chore: remove stresstest/infer.py (deleted in upstream main)
amathxbt
committed
Merge main into fix branch: rebased llm.py with all PR changes
amathxbt
committed
Merge main into fix branch: rebased llm_test.py with all PR test additions
amathxbt
committed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

fix: 5 critical bugs — TEE randomization, HTTP 402 hint, int dtype, inference None crash, gas estimation#201

fix: 5 critical bugs — TEE randomization, HTTP 402 hint, int dtype, inference None crash, gas estimation#201
amathxbt wants to merge 27 commits intoOpenGradient:mainfrom
amathxbt:fix/critical-5-bugs-tee-402-dtype-inference-gas

Commits on Mar 23, 2026

fix(tee_registry): randomize TEE selection to improve load distribution

fix(llm): surface HTTP 402 as actionable error with OPG approval hint

fix(_conversions): return int for zero-decimal fixed-point values

fix(alpha): guard against None from inference node + use estimate_gas in run_workflow

Commits on Mar 26, 2026

fix: make convert_fixed_point_to_python type-stable (always np.float32), replace literal tabs with \t

fix: separate ContractLogicError from generic Exception in gas estimation fallback

test: fix TestGetLlmTee to patch random.choice for deterministic CI

fix: merge main into llm.py — restore _connect_tee, _refresh_tee, _call_with_tee_retry; keep 402 hint, json alias, OPG 0.05

Commits on Mar 27, 2026

test: add HTTP 402 hint coverage for completion, chat, and streaming

fix: merge main into llm_test.py — restore retry/SSL tests, add 402 hint coverage, fix mock status_code

fix: remove stale Union import from _conversions.py (convert_fixed_point_to_python now always returns np.float32)

chore: sync .github/workflows/release.yml from upstream main

chore: sync README.md from upstream main

chore: sync WINDOWS_INSTALL.md from upstream main

chore: sync docs/opengradient/client/llm.md from upstream main

chore: sync docs/opengradient/index.md from upstream main

chore: sync integrationtest/llm/test_llm_chat.py from upstream main

chore: sync pyproject.toml from upstream main

chore: sync src/opengradient/client/_utils.py from upstream main

chore: sync src/opengradient/client/opg_token.py from upstream main

chore: sync tests/opg_token_test.py from upstream main

chore: sync stresstest/llm.py from upstream main

chore: sync stresstest/utils.py from upstream main

chore: sync uv.lock from upstream main

chore: remove stresstest/infer.py (deleted in upstream main)

Merge main into fix branch: rebased llm.py with all PR changes

Merge main into fix branch: rebased llm_test.py with all PR test additions