ngxson/llama.cpp

Pull Requests Commits

fix setting prob for sampled token

ngxson committed 1 year ago

5b966df1

correct comment placement

ngxson committed 1 year ago

a217382b

rename struct token_prob to prob_info

ngxson committed 1 year ago

65ef1c8d

resolve review comments

ngxson committed 1 year ago

d2463dc8

"top_probs" with "post_sampling_probs"

ngxson committed 1 year ago

fd4cf34b

remove --multi-token-probs

ngxson committed 1 year ago

8734df73

update docs [no ci]

ngxson committed 1 year ago

75fe7751

add `post_sampling_probs` option

ngxson committed 1 year ago

ecadd37c

Merge branch 'master' into xsn/fix_logprobs

ngxson committed 1 year ago

c0cca53d

server : output embeddings for all tokens when pooling = none (#10861)

ggerganov committed 1 year ago

Verified 152610ed

server : add "tokens" output (#10853)

ggerganov committed 1 year ago

Verified 0e70ba68

server : (embeddings) using same format for "input" and "content" (#10872)

ngxson committed 1 year ago

Verified 46828872

docs: Fix HIP (née hipBLAS) in README (#10880)

brianredbeard committed 1 year ago

Verified 6b064c92

Revert "llama : add Falcon3 support (#10864)" (#10876)

slaren committed 1 year ago

Verified 4da69d1a

Use model->gguf_kv for loading the template instead of using the C API. (#10868)

dranger003 committed 1 year ago

Verified d62b532c

tests: add tests for GGUF (#10830)

JohannesGaessler committed 1 year ago

Verified 081b29bd

ggerganov committed 1 year ago

Verified 5437d4aa

cmake : fix "amd64" processor string (whisper/2638)

ggerganov committed 1 year ago

Verified 78f76676

vulkan : fix soft_max.comp division by zero (whisper/2633)

gn64 committed 1 year ago

Verified 8dd19a48

ggml : remove return from ggml_gallocr_allocate_node (ggml/1048)

danbev committed 1 year ago

Verified 130d0c90

ggml : add check for grad_accs (ggml/1046)

danbev committed 1 year ago

Verified 3919da8e

ggml : update ggml_backend_cpu_device_supports_op (#10867)

ggerganov committed 1 year ago

Verified 0006f5a7

server : fill usage info in embeddings and rerank responses (#10852)

krystiancha committed 1 year ago

Verified 05c3a444

llama : add Falcon3 support (#10864)

mokeddembillel committed 1 year ago

Verified 382bc7f2

readme : update typos (#10863)

ruanych committed 1 year ago

Verified 4f51968a

server : (UI) fix missing async generator on safari (#10857)

ngxson committed 1 year ago

Verified 227d7c5a

vulkan: bugfixes for small subgroup size systems + llvmpipe test (#10809)

netrunnereve committed 1 year ago

Verified 7b1ec53f

rwkv6: add wkv6 support for Vulkan backend (#10829)

zhiyuan1i committed 1 year ago

Verified 160bc039

unicode : improve naming style (#10838)

ggerganov committed 1 year ago

Verified 08ea539d

sampling : refactor + optimize penalties sampler (#10803)

ggerganov committed 1 year ago

Verified 644fd71b

Older