Morph MORPH
SIGN IN SIGN UP
dhruvhead / tpu-inference UNCLAIMED
0 0 1 Python

COMMITS

main
August 27, 2026
M
bench: dual write benchmark results to BigQuery (#3472)
Ming Huang committed
5e2c712
V
[skip ci] Update nightly support matrices (v6e/v7x)
vllm-ci-bot[bot] committed
c5c8a05
M
bench: port the bm-infra v6e sonnet workloads to daily Buildkite cases (#3465)
Ming Huang committed
8791f57
August 26, 2026
Y
Re-land: bit-pack GDN v3 per-p_id metadata (#3411)
Yuhang Ning (Daniel) committed
9c6037c
V
[skip ci] Update nightly support matrices for vllm (v6e/v7x)
vllm-ci-bot[bot] committed
a7534e2
V
[skip ci] Update nightly support matrices (v6e/v7x)
vllm-ci-bot[bot] committed
0d57ca6
August 25, 2026
M
Remove explicit PAT from pipeline now that we have Github App (#3054)
Ming Huang committed
5f999f0
B
[skip ci] Update nightly support matrices for flax_nnx (v6e/v7x)
Buildkite Bot committed
d896ddd
W
[PCP] Remove unnecessary copy op before cache phase (#3459)
wenxindongwork committed
f5eb3ab
Q
[Bugfix] Compile the Qwen2.5-VL vision tower on the torchax path (mm-encoder JIT padding + ViT pad-segment fixes) (#3450)
Qiliang Cui committed
980abe5
Q
[skip-ci][CODEOWNERS] Add @jrplatin as an owner of /tests/layers/ (#3460)
Qiliang Cui committed
89b65e2
August 24, 2026
Q
[CI] Mark Qwen2.5-VL-7B benchmarks as unverified on the vllm (torchax) nightly variant (#3446)
Qiliang Cui committed
62ad6a9
B
Update verified commit hashes
Buildkite Bot committed
ab333c5
Q
[CI] Auto-retry DCN P/D disagg steps once on transient TPU session init failure (#3448)
Qiliang Cui committed
73de2b6
N
Add warning comment to model loader log (#3433)
Noa Wiesel committed
98079c9
B
[skip ci] Update nightly support matrices for flax_nnx (v6e/v7x)
Buildkite Bot committed
822b8f0
B
[skip ci] Update nightly support matrices for vllm (v6e/v7x)
Buildkite Bot committed
f26e011
B
[skip ci] Update nightly support matrices (v6e/v7x)
Buildkite Bot committed
1942901
August 23, 2026
Q
[CI] Fail the mm benchmark when it produced no decoded tokens (#3447)
Qiliang Cui committed
b98d381
B
Update verified commit hashes
Buildkite Bot committed
78da877
August 22, 2026
B
Update verified commit hashes
Buildkite Bot committed
dacd27d
B
[skip ci] Update nightly support matrices for vllm (v6e/v7x)
Buildkite Bot committed
b8692ff
H
[tpu-inference] Add block-diffusion strategy configuration (#3242)
hx committed
3582484
B
Update verified commit hashes
Buildkite Bot committed
7482cf9
B
[skip ci] Update nightly support matrices (v6e/v7x)
Buildkite Bot committed
0fbf23a
X
Use tokamax release version to unblock TPU-inference release. (#3441)
XiongfeiWei committed
d253b79
August 21, 2026
W
[Fix] Reduce shared-expert output over in-group TP under attention DP (#3435)
wenxindongwork committed
57987c2
Q
[gemma4] Fix torchax-path E2B/E4B accuracy: inline PLE + skip KV-cache writes for KV-shared layers (#3406)
Qiliang Cui committed
ee1e17a
B
[skip ci] Update nightly support matrices for vllm (v6e/v7x)
Buildkite Bot committed
07b666a
B
[skip ci] Update nightly support matrices (v6e/v7x)
Buildkite Bot committed
923bbe4