tqcq air

tqcq synced new reference refs/tags/b10080 to ggml-org/llama.cpp from mirror 2026-07-22 20:09:37 +08:00

tqcq synced commits to refs/heads/master at ggml-org/llama.cpp from mirror 2026-07-22 20:09:37 +08:00

6d5a910c50 common: infer the speculative type from the draft repo sidecars (#25989)

f534da26e4 Fix DeepSeek4 crafted template (#25414)

3ce7da2c85 ggml: enable PowerPC backend variants on AIX (#25983)

b4d6c7d8ff ci : fix SYCL package shared library lookup (#25987)

7347430f44 webgpu : add CONV_2D_DW (depthwise conv2d) kernel (#25847)

Compare 6 commits »

tqcq synced commits to refs/tags/b10088 at ggml-org/llama.cpp from mirror 2026-07-22 20:09:37 +08:00

tqcq synced new reference refs/tags/b10088 to ggml-org/llama.cpp from mirror 2026-07-22 20:09:37 +08:00

tqcq synced commits to refs/tags/b10087 at ggml-org/llama.cpp from mirror 2026-07-22 20:09:37 +08:00

tqcq synced new reference refs/tags/b10087 to ggml-org/llama.cpp from mirror 2026-07-22 20:09:37 +08:00

tqcq synced commits to refs/tags/b10085 at ggml-org/llama.cpp from mirror 2026-07-22 20:09:37 +08:00

tqcq synced new reference refs/tags/b10085 to ggml-org/llama.cpp from mirror 2026-07-22 20:09:37 +08:00

tqcq synced commits to refs/tags/b10084 at ggml-org/llama.cpp from mirror 2026-07-22 20:09:37 +08:00

tqcq synced commits to refs/tags/b10080 at ggml-org/llama.cpp from mirror 2026-07-22 20:09:37 +08:00

tqcq synced new reference refs/tags/b10077 to ggml-org/llama.cpp from mirror 2026-07-22 11:59:34 +08:00

tqcq synced commits to refs/tags/b10078 at ggml-org/llama.cpp from mirror 2026-07-22 11:59:34 +08:00

tqcq synced new reference refs/tags/b10078 to ggml-org/llama.cpp from mirror 2026-07-22 11:59:34 +08:00

tqcq synced commits to refs/tags/b10077 at ggml-org/llama.cpp from mirror 2026-07-22 11:59:34 +08:00

tqcq synced commits to refs/heads/master at ggml-org/llama.cpp from mirror 2026-07-22 11:59:34 +08:00

67b9b0e7f6 llama-arch: fix DeepSeek4 APE tensor op (#25945)

1f66c3ce1c Add support for Laguna XS.2 & M.1 (#25165)

66e4bf7e59 convert: fix handle HunyuanVL XD-RoPE config (#25514)

b4aa7dd477 mtmd : use align_corners for qwen3vl vision position embedding interpolation (#25781)

71102a73f2 hexagon: check tensor type when reusing descriptors (#25968)

Compare 5 commits »

tqcq synced commits to refs/tags/b10079 at ggml-org/llama.cpp from mirror 2026-07-22 11:59:34 +08:00

tqcq synced new reference refs/tags/b10079 to ggml-org/llama.cpp from mirror 2026-07-22 11:59:34 +08:00

tqcq synced commits to refs/tags/b10076 at ggml-org/llama.cpp from mirror 2026-07-22 03:49:34 +08:00

tqcq synced new reference refs/tags/b10076 to ggml-org/llama.cpp from mirror 2026-07-22 03:49:34 +08:00

tqcq synced commits to refs/heads/master at ggml-org/llama.cpp from mirror 2026-07-22 03:49:34 +08:00

846e991ec3 cuda: add sqrt_softplus in topk-moe for dsv4 (#25896)

fb0e6b6219 kleidiai : warn once when a weight type has no KleidiAI kernel (#25701)

60f6a17704 common: resolve draft repo to its requested sidecar (#25955)

fd41bf65a2 server: return 400 instead of 500 on validation error with X-Conversation-Id (#25760)

40b740ad05 server : properly handle null llama_context (#25868)

Compare 8 commits »