HappyZ

happyz synced commits to refs/pull/18968/merge at happyz/llama.cpp from mirror 2026-02-06 18:02:09 -08:00

cc58ade7c0 Merge 01adb243e4 into 7fbd36c50c

7fbd36c50c ggml-webgpu: JIT compile binary operators and handle binding overlaps (#19310)

537eadb1b9 sycl: add F16 support for GGML_OP_CEIL (#19306)

db6adb3c88 tests: reduce number of FA test permutations (#19381)

dfde5993ea common : add common_speculative_is_compat() (#19270)

Compare 6 commits »

happyz synced commits to refs/pull/18862/merge at happyz/llama.cpp from mirror 2026-02-06 18:02:08 -08:00

a078f09648 Merge 6693df846c into b83111815e

b83111815e model : support Step3.5-Flash (#19283)

3228e77287 gguf-py : bump sentencepiece version (#19319)

7fbd36c50c ggml-webgpu: JIT compile binary operators and handle binding overlaps (#19310)

537eadb1b9 sycl: add F16 support for GGML_OP_CEIL (#19306)

Compare 8 commits »

happyz synced commits to refs/pull/18858/merge at happyz/llama.cpp from mirror 2026-02-06 18:02:08 -08:00

f7887cd9f0 Merge a73b9d361a into 06bf3796f4

06bf3796f4 unicode : MSVC regex fix (#19340)

3688c4f504 Kimi-Linear support (backend agnostic + MLA KV cache) (#18755)

1946e46f4c vulkan: For coopmat2 FA, use fp16 accumulators for the final result (#19376)

f9bd518a6b vulkan: make FA mask/softcap enables spec constants (#19309)

Compare 12 commits »

happyz synced commits to refs/pull/18886/head at happyz/llama.cpp from mirror 2026-02-06 18:02:08 -08:00

64f05859db add llama_context graph_type

a2860dc85e Merge branch 'master' into xsn/mtp_model

b83111815e model : support Step3.5-Flash (#19283)

3228e77287 gguf-py : bump sentencepiece version (#19319)

7fbd36c50c ggml-webgpu: JIT compile binary operators and handle binding overlaps (#19310)

Compare 209 commits »

happyz synced commits to refs/pull/18886/merge at happyz/llama.cpp from mirror 2026-02-06 18:02:08 -08:00

dd6582b7e6 Merge 64f05859db into b83111815e

64f05859db add llama_context graph_type

a2860dc85e Merge branch 'master' into xsn/mtp_model

b83111815e model : support Step3.5-Flash (#19283)

3228e77287 gguf-py : bump sentencepiece version (#19319)

Compare 11 commits »

happyz synced commits to refs/pull/18816/merge at happyz/llama.cpp from mirror 2026-02-06 18:02:07 -08:00

bb5ccf8231 Merge 6cf18ecd6b into b83111815e

b83111815e model : support Step3.5-Flash (#19283)

3228e77287 gguf-py : bump sentencepiece version (#19319)

7fbd36c50c ggml-webgpu: JIT compile binary operators and handle binding overlaps (#19310)

537eadb1b9 sycl: add F16 support for GGML_OP_CEIL (#19306)

Compare 8 commits »

happyz synced commits to refs/pull/18825/merge at happyz/llama.cpp from mirror 2026-02-06 18:02:07 -08:00

6e7a78e666 Merge f8b30aa31c into 3228e77287

3228e77287 gguf-py : bump sentencepiece version (#19319)

7fbd36c50c ggml-webgpu: JIT compile binary operators and handle binding overlaps (#19310)

537eadb1b9 sycl: add F16 support for GGML_OP_CEIL (#19306)

db6adb3c88 tests: reduce number of FA test permutations (#19381)

Compare 6 commits »

happyz synced commits to refs/pull/18756/merge at happyz/llama.cpp from mirror 2026-02-06 18:02:07 -08:00

2f25588634 Merge f3bce52986 into db6adb3c88

db6adb3c88 tests: reduce number of FA test permutations (#19381)

dfde5993ea common : add common_speculative_is_compat() (#19270)

06bf3796f4 unicode : MSVC regex fix (#19340)

3688c4f504 Kimi-Linear support (backend agnostic + MLA KV cache) (#18755)

Compare 20 commits »

happyz synced commits to refs/pull/18792/head at happyz/llama.cpp from mirror 2026-02-06 18:02:07 -08:00

ae33204660 Fix bad permute

6d0a9adace Change to decay mask approach

fb80f6e370 Adapt autoregressive version from @ymcki

7317889cd4 Refactor and optimize

7399880f12 Remove old methods.

Compare 10 commits »

happyz synced commits to refs/pull/18792/merge at happyz/llama.cpp from mirror 2026-02-06 18:02:07 -08:00

eab7a4a330 Merge ae33204660 into b83111815e

b83111815e model : support Step3.5-Flash (#19283)

3228e77287 gguf-py : bump sentencepiece version (#19319)

7fbd36c50c ggml-webgpu: JIT compile binary operators and handle binding overlaps (#19310)

ae33204660 Fix bad permute

Compare 13 commits »

happyz synced commits to refs/pull/18675/merge at happyz/llama.cpp from mirror 2026-02-06 18:02:06 -08:00

198b3998d2 Merge c730c343de into b83111815e

b83111815e model : support Step3.5-Flash (#19283)

3228e77287 gguf-py : bump sentencepiece version (#19319)

7fbd36c50c ggml-webgpu: JIT compile binary operators and handle binding overlaps (#19310)

537eadb1b9 sycl: add F16 support for GGML_OP_CEIL (#19306)

Compare 8 commits »

happyz synced commits to refs/pull/18655/merge at happyz/llama.cpp from mirror 2026-02-06 18:02:06 -08:00

cbb8416178 Merge ffa26323d8 into b83111815e

ffa26323d8 Merge remote-tracking branch 'ggml-org/master' into allozaur/mcp-mvp

c50f191a0d chore: update webui build output

ccadd26024 fix: Syntax highlight for `svelte` and `vue`

3287a41af9 feat: MCP Resources

Compare 153 commits »

happyz synced commits to refs/pull/18745/merge at happyz/llama.cpp from mirror 2026-02-06 18:02:06 -08:00

2dcd2b64ff Merge d7ec22fe46 into 537eadb1b9

537eadb1b9 sycl: add F16 support for GGML_OP_CEIL (#19306)

db6adb3c88 tests: reduce number of FA test permutations (#19381)

dfde5993ea common : add common_speculative_is_compat() (#19270)

06bf3796f4 unicode : MSVC regex fix (#19340)

Compare 6 commits »

happyz synced commits to refs/pull/18655/head at happyz/llama.cpp from mirror 2026-02-06 18:02:05 -08:00

ffa26323d8 Merge remote-tracking branch 'ggml-org/master' into allozaur/mcp-mvp

c50f191a0d chore: update webui build output

ccadd26024 fix: Syntax highlight for `svelte` and `vue`

3287a41af9 feat: MCP Resources

b83111815e model : support Step3.5-Flash (#19283)

Compare 98 commits »

happyz synced commits to refs/pull/18604/merge at happyz/llama.cpp from mirror 2026-02-06 18:02:04 -08:00

de36a357a8 Merge 43d9e59d0a into 06bf3796f4

06bf3796f4 unicode : MSVC regex fix (#19340)

3688c4f504 Kimi-Linear support (backend agnostic + MLA KV cache) (#18755)

1946e46f4c vulkan: For coopmat2 FA, use fp16 accumulators for the final result (#19376)

f9bd518a6b vulkan: make FA mask/softcap enables spec constants (#19309)

Compare 18 commits »

happyz synced commits to refs/pull/18626/merge at happyz/llama.cpp from mirror 2026-02-06 18:02:04 -08:00

bf10c64cdd Merge 9eec24ad8a into b83111815e

b83111815e model : support Step3.5-Flash (#19283)

3228e77287 gguf-py : bump sentencepiece version (#19319)

7fbd36c50c ggml-webgpu: JIT compile binary operators and handle binding overlaps (#19310)

537eadb1b9 sycl: add F16 support for GGML_OP_CEIL (#19306)

Compare 8 commits »

happyz synced commits to refs/pull/18633/merge at happyz/llama.cpp from mirror 2026-02-06 18:02:04 -08:00

d58a1c0779 Merge 1f0d90c3dd into 537eadb1b9

537eadb1b9 sycl: add F16 support for GGML_OP_CEIL (#19306)

db6adb3c88 tests: reduce number of FA test permutations (#19381)

dfde5993ea common : add common_speculative_is_compat() (#19270)

06bf3796f4 unicode : MSVC regex fix (#19340)

Compare 14 commits »

happyz synced commits to refs/pull/18493/merge at happyz/llama.cpp from mirror 2026-02-06 18:02:03 -08:00

f02c3ce692 Merge eb81f41889 into 7fbd36c50c

7fbd36c50c ggml-webgpu: JIT compile binary operators and handle binding overlaps (#19310)

537eadb1b9 sycl: add F16 support for GGML_OP_CEIL (#19306)

db6adb3c88 tests: reduce number of FA test permutations (#19381)

dfde5993ea common : add common_speculative_is_compat() (#19270)

Compare 5 commits »

happyz synced commits to refs/pull/18574/merge at happyz/llama.cpp from mirror 2026-02-06 18:02:03 -08:00

9b249554a3 Merge 4ac643ee74 into 537eadb1b9

537eadb1b9 sycl: add F16 support for GGML_OP_CEIL (#19306)

db6adb3c88 tests: reduce number of FA test permutations (#19381)

dfde5993ea common : add common_speculative_is_compat() (#19270)

06bf3796f4 unicode : MSVC regex fix (#19340)

Compare 12 commits »

happyz synced commits to refs/pull/18586/merge at happyz/llama.cpp from mirror 2026-02-06 18:02:03 -08:00

ab23f84a1b Merge 646e4d0c85 into db6adb3c88

db6adb3c88 tests: reduce number of FA test permutations (#19381)

dfde5993ea common : add common_speculative_is_compat() (#19270)

06bf3796f4 unicode : MSVC regex fix (#19340)

3688c4f504 Kimi-Linear support (backend agnostic + MLA KV cache) (#18755)

Compare 19 commits »