ochafik/llama.cpp

Pull Requests Commits

llama2.c converter: cleanups + take n_ff from config

ochafik committed 2 years ago

c58792c7

llama2.c: escape whitespaces w/ U+2581 in vocab converter the llama.cpp way

ochafik committed 2 years ago

0722e58a

llama2.c: use defines for gguf keys

ochafik committed 2 years ago

20c44711

llama2.c: update default path for vocab model + readme

ochafik committed 2 years ago

df3b81ab

llama2.c: support copying vocab from a llama gguf model file

ochafik committed 2 years ago

c88362d1

llama2.c gguf conversion: fix token types in converter

ochafik committed 2 years ago

63174b80

Simplify vector building logic

ochafik committed 2 years ago

463e1178

llama2.c: direct gguf output (WIP)

ochafik committed 2 years ago

5132130a

llm : add Falcon support (#2717)

ggerganov committed 2 years ago

Verified cf658adc

minor : fix trailing whitespace

ggerganov committed 2 years ago

Verified a192860c

examples : restore the functionality to import llama2.c models (#2685)

ochafik committed 2 years ago

Verified 95385241

fix convert-lora-to-ggml.py (#2738)

slaren committed 2 years ago

Verified 335acd2f

main : insert bos if no tokens (#2727)

klosax committed 2 years ago

Verified 5290c38e

gitignore : fix for windows (#2729)

akawrykow committed 2 years ago

Verified cc34dbda

chmod : make scripts executable (#2675)

cebtenzzre committed 2 years ago

Verified 7c2227a1

devops : RPM Specs (#2723)

jboero committed 2 years ago

Verified f19dca04

Fix values shown in the quantize tool help (#2735)

ikawrakow committed 2 years ago

Verified 8207214b

Strided perplexity (#2714)

ikawrakow committed 2 years ago

Verified 62959e74

Fix ggml to gguf conversion on Windows (#2733)

IgnacioFDM committed 2 years ago

Verified 7f7ddd50

server : allow json array in prompt or content for direct token input (#2306)

jxy committed 2 years ago

Verified b8ad1b66

docs : add grammar docs (#2701)

ejones committed 2 years ago

Verified f5fe98d1

Improve handling of special tokens in GGML to GGUF converter (#2725)

KerfuffleV2 committed 2 years ago

Verified 777f42ba

llama : fix whitespace escaping in tokenizer (#2724)

goerch committed 2 years ago

Verified 46ef5b5f

CUDA: use mul_mat_q kernels by default (#2683)

JohannesGaessler committed 2 years ago

Verified c63bb1d1

convert.py : clarifying error message (#2718)

apetenchea committed 2 years ago

Verified 3b6cfe7c

Fix CUDA softmax by subtracting max value before exp (#2665)

li-plus committed 2 years ago

Verified 800c9635

gguf : add ftype meta info to the model (#2710)

ggerganov committed 2 years ago

Verified deb7dfca

Quantization imrovements for k_quants (#2707)

ikawrakow committed 2 years ago

Verified bac66994

embedding : evaluate prompt in batches (#2713)

slaren committed 2 years ago

Verified 519c981f

ggml-cuda : use graph allocator (#2684)

slaren committed 2 years ago

Verified 1123f7fb

Older