ggerganov/llama.cpp

Pull Requests Commits

ggml : force F32 precision for ggml_mul_mat

ggerganov committed 1 year ago

Verified a40f6110

ggml-cuda: Fix HIP build (#4528)

arlo-phoenix committed 1 year ago

Verified a7aee47b

llama.swiftui : add tinyllama 1.1B F16

ggerganov committed 1 year ago

Verified 0e18b2e7

llama.swiftui : add more models

ggerganov committed 1 year ago

Verified 6ff39b12

llama : add phi-2 + fix NeoX rope + ggml_mul_mat_set_prec (#4490)

ebeyabraham committed 1 year ago

Verified b9e74f9b

llama : fix try_override for bool_value which always return true (#4519)

hankcs committed 1 year ago

Verified 3c04bf6d

decode : fix logits_valid for legacy API (#4516)

cebtenzzre committed 1 year ago

Verified 2994f0c5

readme : update hot topics

ggerganov committed 1 year ago

Verified b1306c43

llama.swiftui : add bench functionality (#4483)

ggerganov committed 1 year ago

Verified 800a489e

gguf-py : fail fast on nonsensical special token IDs (#4489)

cebtenzzre committed 1 year ago

Verified f7f468a9

build : Check the ROCm installation location (#4485)

Su3h7aM committed 1 year ago

Verified 919c4066

finetune : keep allocs alive until all allocations are done (#4486)

slaren committed 1 year ago

Verified 45668633

server : disable llm logs if SERVER_VERBOSE is off (#3792)

olexiyb committed 1 year ago

Verified 0ffc92d2

server : fix grammar being ignored (#4494)

AdithyanI committed 1 year ago

Verified 8edd2b40

server : fix possible ambiguity in content type charset (#4501)

z80maniac committed 1 year ago

Verified eb16dae7

server : allow requests larger than 8K (#4500)

mzcu committed 1 year ago

Verified 62bd52b7

Link to cublas dynamically on Windows even with LLAMA_STATIC (#4506)

bullno1 committed 1 year ago

Verified 5daa5f54

lora : add support for non-llama models (#3333)

slaren committed 1 year ago

Verified c6c4fc08

llama : sanity checks for access to logits (#4274)

cebtenzzre committed 1 year ago

Verified 8a5be3bd

server : add optional API Key Authentication example (#4441)

ShadovvBeast committed 1 year ago

Verified 88ae8952

ggml : group mul_mat_id rows by matrix (cpu only) (#4480)

slaren committed 1 year ago

Verified ee4725a6

ggml : use ggml_row_size where possible (#4472)

slaren committed 1 year ago

Verified 6744dbe9

ggml : remove n_dims from ggml_tensor (#4469)

slaren committed 1 year ago

Verified cafcd4f8

py : add protobuf dependency (#4466)

strutive07 committed 1 year ago

Verified c50e4001

ggml : add ggml_row_size() (fixes llama out of space) (#4461)

LostRuins committed 1 year ago

Verified 20a68a70

ggml : fix OpenCL broadcast requirement for ggml_mul (close #4453)

ggerganov committed 1 year ago

Verified 55e87c37

convert : support loading vocab from fast tokenizer config (#3633)

strutive07 committed 1 year ago

Verified 873637af

readme : update supported model list (#4457)

BarfingLemurs committed 1 year ago

Verified 0353a184

server : fix handling of characters that span multiple tokens when streaming (#4446)

shibe2 committed 1 year ago

Verified 948ff137

sync : ggml (SD ops, tests, kernels) (#4444)

ggerganov committed 1 year ago

Verified 4d98d9a6

Older