 Kawrakow
|
7051aacfac
winogrande: evaluate log-probs in parallel (#5036)
|
2 rokov pred |
 chiranko
|
2b3b999cac
llama : add CodeShell support (#5016)
|
2 rokov pred |
|
Kawrakow
|
993fba8180
perplexity: avoid unnecessary alloocations and logit copies (#5035)
|
2 rokov pred |
 Georgi Gerganov
|
8b20858e5e
perplexity : faster Winogrande via batching (#5024)
|
2 rokov pred |
 John
|
57e2a7a52a
llama : fix falcon arch for tied output embeddings (#4978)
|
2 rokov pred |
|
Georgi Gerganov
|
9b6ea4263a
cmake : add ggml public headers (#5011)
|
2 rokov pred |
 Xuan Son Nguyen
|
821f0a271e
server : defer tasks when "slot unavailable" (#5018)
|
2 rokov pred |
 slaren
|
96d7f56d29
llama : fix mlock with no-mmap with Metal (#5025)
|
2 rokov pred |
|
Georgi Gerganov
|
2d5419d08a
imatrix : fix assert for src0 non-cont check
|
2 rokov pred |
|
Georgi Gerganov
|
d391ae9b49
perplexity : fix winogrande N tasks option
|
2 rokov pred |
|
Georgi Gerganov
|
e9240cdfa0
scripts : add get-winogrande.sh
|
2 rokov pred |
 David Sommers
|
b46757735d
convert.py : fix llama/llama2 conversion due to vocab_size=-1 (#5019)
|
2 rokov pred |
|
Kawrakow
|
3e945cc1e9
HellaSwag: speed up by parallelizing log-prob evaluation (#5020)
|
2 rokov pred |
|
Georgi Gerganov
|
ad19812cda
perplexity : faster HellaSwag via batching (#5017)
|
2 rokov pred |
|
Kawrakow
|
682986a08e
Add Winogrande evaluation (#5015)
|
2 rokov pred |
|
Georgi Gerganov
|
dcad445d0c
scritps : add helper script to get hellaswag data in txt format
|
2 rokov pred |
 Paul Tsochantaris
|
1e605f4102
metal : fix memory leak, dangling pointer and unused autorel (#5007)
|
2 rokov pred |
|
Georgi Gerganov
|
6b6916b215
sync : ggml
|
2 rokov pred |
|
Georgi Gerganov
|
38566680cd
ggml : add IQ2 to test-backend-ops + refactoring (#4990)
|
2 rokov pred |
|
Georgi Gerganov
|
ba69bbc84c
imatrix : offload to GPU support (#4957)
|
2 rokov pred |
|
Georgi Gerganov
|
44a1a4a41a
backend : add eval callback (#4935)
|
2 rokov pred |
|
Georgi Gerganov
|
c918fe8dca
metal : create autorelease pool during library build (#4970)
|
2 rokov pred |
|
Georgi Gerganov
|
0f83e727af
py : fix whitespace
|
2 rokov pred |
|
Georgi Gerganov
|
4f4bf35f46
py : fix missing added_tokens_dict for SPM and BPE vocabs (#4971)
|
2 rokov pred |
|
Kawrakow
|
2b3a665d39
llama : use Q4_K for attn_v for Q2_K_S when n_gqa >= 4 (#4996)
|
2 rokov pred |
|
Paul Tsochantaris
|
7563293665
metal : remove unnecessary nil check (#4986)
|
2 rokov pred |
 David Renshaw
|
f46c0c1b0e
llama : fix copy/paste error in llama_sampling_params comment (#4994)
|
2 rokov pred |
|
Georgi Gerganov
|
5c99960901
py : remove unnecessary hasattr (#4903)
|
2 rokov pred |
 Philip Taron
|
bee938da74
nix: remove nixConfig from flake.nix (#4984)
|
2 rokov pred |
 Daniel Bevenius
|
cec8a48470
finetune : add training data file to log message (#4979)
|
2 rokov pred |
