ggerganov/llama.cpp

Pull Requests Commits

vulkan: optimizations for deepseek prompt processing (#14555)

jeffbolznv committed 91 days ago

Verified 98197e5c

model : support LiquidAI LFM2 hybrid family (#14620)

tdakhran committed 91 days ago

Verified f5e96b36

HIP : Add HIP 7.0+ compatibility for hipBLAS compute types (#14634)

slojosic-amd committed 91 days ago

Verified 756aa102

readme : add hot PRs (#14636)

ggerganov committed 91 days ago

Verified aaa088d8

llama : move enum llama_vocab_pre_type to implementation (#14631)

ggerganov committed 91 days ago

Verified 0d5375d5

vocab : add midm-2.0 model pre-tokenizer (#14626)

Bing-su committed 92 days ago

Verified 576c82ed

model : Granite Four (#13550)

gabe-l-hart committed 92 days ago

Verified 0aedae00

opencl: add tiled mul_mat_f16_f32 (#14535)

rmatif committed 92 days ago

Verified 6bdda139

opencl: add `set_rows` for `f16` and `f32` (#14547)

lhez committed 92 days ago

Verified 0b885577

Smoldocling support (#14597)

ryan-mangeno committed 92 days ago

Verified 4bb625b7

Docs: script to auto-generate ggml operations docs (#14598)

am17an committed 92 days ago

Verified 11ee0fea

cmake : do not search for curl libraries by ourselves (#14613)

EZForever committed 92 days ago

Verified a4575513

SYCL: Initial set_rows kernel implementation (#14562)

qnixsynapse committed 93 days ago

Verified 704bb7a7

llama : minor coding style fix for smollm3 (#14605)

ngxson committed 93 days ago

Verified 435a6d10

cmake : bump llguidance version to v1.0.1 (#14609)

EZForever committed 93 days ago

Verified f9a867f5

cmake : llguidance build parser library only (#14608)

EZForever committed 93 days ago

Verified ac44eb6c

cuda : support Falcon-H1 state size for SSM_SCAN (#14602)

compilade committed 93 days ago

Verified a57d1bcb

llama : remove llm_graph_input_one (#14603)

ngxson committed 93 days ago

Verified cb9178f8

llama : support Jamba hybrid Transformer-Mamba models (#7531)

compilade committed 93 days ago

Verified 4a5686da

ggml : add ggml_scale_bias (#14417)

ngxson committed 93 days ago

Verified 98bab638

ggml : prevent integer overflow in gguf tensor size calculation (#14595)

Yuuoniy committed 93 days ago

Verified 26a48ad6

model : add skt/A.X-4.0 model vocabulary (#14589)

Bing-su committed 94 days ago

Verified ffd59e7d

llama : remove unintended whitespace (#14592)

CISC committed 94 days ago

Verified 10555459

model : add support for Falcon-H1 family (#14534)

ibrahimkhadraoui committed 94 days ago

Verified 04655063

convert : fix smollm3 jinja template (#14586)

ngxson committed 94 days ago

Verified 20b7bf8a

vulkan: optimize flash attention split_k_reduce (#14554)

jeffbolznv committed 94 days ago

Verified 6efcd659

model : fix hunyuan moe chat template (#14584)

stevenkuang-tencent committed 94 days ago

Verified 699f4392

model : add SmolLM3 (#14581)

ngxson committed 94 days ago

Verified 08382869

memory : fix broken batch splits for recurrent cache (#14575)

compilade committed 94 days ago

Verified bb4f7a9e

vulkan : fix rope with partial rotation and non-cont src (#14582)

jeffbolznv committed 94 days ago

Verified b8eeb874

Newer Older