Skip to content

Distribute and run LLMs with a single file.

Commits on main

Upgrade to Cosmopolitan v3.7.1
Justine Tunney committed 2 years ago
d0b5e8f
Fix GitHub Actions
Justine Tunney committed 2 years ago
bc7d85c
Increase tokens per second on tiny models
Justine Tunney committed 2 years ago
584a327
Add GeLU LUT code for analysis
Justine Tunney committed 2 years ago
c39d30c
Write explicitly vectorized GeLU functions
Justine Tunney committed 2 years ago
8ace604
Fix more issues
Justine Tunney committed 2 years ago
fd73706
Add seccomp, tokenbucket, and batch prioritization
Justine Tunney committed 2 years ago
99dd1c0
Use ruler reduction for GGML dot products
Justine Tunney committed 2 years ago
5b06924
Add better support for AVX512 CPUs
Justine Tunney committed 2 years ago
45130eb
Make GeLU go 10x faster (take two)
Justine Tunney committed 2 years ago
a4f942f
Fix sync bug in new task pool
Justine Tunney committed 2 years ago
835985b
Work out kinks with the latest sync
Justine Tunney committed 2 years ago
e914f13
Get whisperfile working very well
Justine Tunney committed 2 years ago
5da8d62
Synchronize with llama.cpp upstream
Justine Tunney committed 2 years ago
e9ee3f9
Use ar.ape
Justine Tunney committed 2 years ago
a22dd04
Make GGML threads spawn 10x faster
Justine Tunney committed 2 years ago
cda83f8
Make more fixes to tracing feature
Justine Tunney committed 2 years ago
2699f6b
Fix new tracing feature
Justine Tunney committed 2 years ago
edc3f2d
Make GeLU 10x faster
Justine Tunney committed 2 years ago
2781abf
Prevent rare double printing of stats
Justine Tunney committed 2 years ago
8ba17ea
Add chrome://tracing/ feature
Justine Tunney committed 2 years ago
d451e0e
Make whisperfile go 2x faster on 96 core cpu
Justine Tunney committed 2 years ago
3b653c7
Update supported platform information in README.md
Justine Tunney committed 2 years ago
13e647b
Add README to whisperfile
Justine Tunney committed 2 years ago
0549317
Disable GPU in whisperfile by default
Justine Tunney committed 2 years ago
2841793
Make whisperfile 12% faster in GPU mode
Justine Tunney committed 2 years ago
b3bdc62
Get CUDA and Metal GPU working in whisperfile
Justine Tunney committed 2 years ago
0849f32
Add CLI program to whisperfile
Justine Tunney committed 2 years ago
94e9629
Merge whisperfile into llamafile (#517)
CJ Pais committed 2 years ago
fd891be
Upgrade stable diffusion
Justine Tunney committed 2 years ago
25ceb2c