 Georgi Gerganov
|
959ef0c0df
perplexity : fix kv cache handling for hellaswag (#4981)
|
il y a 2 ans |
 Kerfuffle
|
91f6499393
Respect tokenizer.ggml.add_bos_token value when tokenizing (#4040)
|
il y a 2 ans |
 cebtenzzre
|
b12fa0d1c1
build : link against build info instead of compiling against it (#3879)
|
il y a 2 ans |
|
Kerfuffle
|
6e08281e58
Extend llama_kv_cache_seq_rm to allow matching any sequence (#3843)
|
il y a 2 ans |
 Marcus Dunn
|
5be6c803fa
llama : remove token functions with `context` args in favor of `model` (#3720)
|
il y a 2 ans |
 slaren
|
16bc66d947
llama.cpp : split llama_context_params into model and context params (#3301)
|
il y a 2 ans |
|
Georgi Gerganov
|
ec893798b7
llama : custom attention mask + parallel decoding + no context swaps (#3228)
|
il y a 2 ans |
|
Cebtenzzre
|
8781013ef6
make : restore build-info.h dependency for several targets (#3205)
|
il y a 2 ans |
|
Cebtenzzre
|
e6616cf0db
examples : add compiler version and target to build info (#2998)
|
il y a 2 ans |
|
Cebtenzzre
|
3aefaab9e5
check C++ code with -Wmissing-declarations (#3184)
|
il y a 2 ans |
|
Cebtenzzre
|
e64f5b5578
examples : make n_ctx warning work again (#3066)
|
il y a 2 ans |
|
Cebtenzzre
|
00d62adb79
fix some warnings from gcc and clang-tidy (#3038)
|
il y a 2 ans |
|
Georgi Gerganov
|
e36ecdccc8
build : on Mac OS enable Metal by default (#2901)
|
il y a 2 ans |
 Kawrakow
|
fa3582f509
Tell users attmepting to run perplexity with too few tokens to use more (#2882)
|
il y a 2 ans |
 Johannes Gäßler
|
6b73ef1201
YAML result logging + preset script (#2657)
|
il y a 2 ans |
|
Kawrakow
|
463173a6c0
llama : speedup tokenization (#2831)
|
il y a 2 ans |
|
Georgi Gerganov
|
edd4c14817
llama : more tokenizer fixes (#2810)
|
il y a 2 ans |
|
Kawrakow
|
771551a793
Fix HellaSwag (#2805)
|
il y a 2 ans |
|
Kawrakow
|
d046dcee08
Faster perplexity computation (#2786)
|
il y a 2 ans |
|
Georgi Gerganov
|
cf658adc83
llm : add Falcon support (#2717)
|
il y a 2 ans |
|
Kawrakow
|
62959e740e
Strided perplexity (#2714)
|
il y a 2 ans |
|
Georgi Gerganov
|
6381d4e110
gguf : new file format with flexible meta data (beta) (#2398)
|
il y a 2 ans |
|
Kawrakow
|
cb1c0727bd
HellaSwag: split token evaluation into batches if needed (#2681)
|
il y a 2 ans |
|
Kawrakow
|
5e9ff54a67
More efficient Hellaswag implementation (#2677)
|
il y a 2 ans |
|
Georgi Gerganov
|
e9b12c332e
perplexity : more meaningful ETA number - 2 decimal points
|
il y a 2 ans |
 Borislav Stanimirov
|
ff966e7ca6
build : fix several cast and printf warnings (#2499)
|
il y a 2 ans |
 klosax
|
8a88e5855c
perplexity : add Hellaswag calculation (#2389)
|
il y a 2 ans |
|
klosax
|
b5fe67f8c6
Perplexity: Compute scores correlated to HellaSwag (#2312)
|
il y a 2 ans |
|
Georgi Gerganov
|
d01bccde9f
ci : integrate with ggml-org/ci (#2250)
|
il y a 2 ans |
 Evan Miller
|
5656d10599
mpi : add support for distributed inference via MPI (#2099)
|
il y a 2 ans |
