ggerganov/llama.cpp

Pull Requests Commits

Regroup q4_1 dot addition for better numerics.

blackhole89 committed 2 years ago

4aeee216

Squeeze out about 5% more performance in Q4_1 inference

blackhole89 committed 2 years ago

580991bb

Temporary bump the memory buffer size - hopefully fix issues from 483bab2e

ggerganov committed 2 years ago

Verified 31572d96

Update README.md (#444)

gjmulder committed 2 years ago

Verified f4f5362e

fix instruct mode (#445)

rabidcopy committed 2 years ago

Verified 863f65e2

Properly free llama_context on failure

ggerganov committed 2 years ago

Verified afd220d9

additional optimizations for POWER9 (#454)

classilla committed 2 years ago

Verified 481044d5

Support calling mlock() on loaded model data on Linux and macOS (#453)

comex committed 2 years ago

Verified 563cdc39

Add embedding mode with arg flag. Currently working (#282)

StrikingLoo committed 2 years ago

Verified 8d4a855c

Add link to Roadmap discussion

ggerganov committed 2 years ago

Verified b6b268d4

Revert "Fix memory allocation issues and seg faults"

ggerganov committed 2 years ago

3cd8dde0

Fix memory allocation issues and seg faults

ggerganov committed 2 years ago

Verified 4870e455

Avoid the transposed X branch in the Z = X * Y matrix multiplication (#439)

ggerganov committed 2 years ago

Verified 483bab2e

Fix quantize script not finding models in parent directory (#428)

j-f1 committed 2 years ago

Verified 404e1da3

Remove oboslete command from Docker script

ggerganov committed 2 years ago

Verified 4cc053b6

ggerganov committed 2 years ago

Verified 0ba5a3a9

Replace EOS with newline to prevent context/memory being flushed by EOS in interactive mode (#333)

rabidcopy committed 2 years ago

Verified 2e17dfd8

Fix GPTQ converter (#423)

R2D2FISH committed 2 years ago

Verified 20a1a4e0

Generate library with CMake (#430)

nusu-github committed 2 years ago

Verified ad072fc5

Command line args bounds checking (#424)

anzz1 committed 2 years ago

Verified ea10d3de

siraben committed 2 years ago

a18c1925

Revert "Delete SHA256SUMS for now" (#429)

sw committed 2 years ago

Verified a50e39c6

Fix Makefile echo escape codes (by removing them). (#418)

KerfuffleV2 committed 2 years ago

Verified a140219e

Move model section from issue template to README.md (#421)

gjmulder committed 2 years ago

Verified 8a3e5ef8

Delete SHA256SUMS for now (#416)

anzz1 committed 2 years ago

Verified 8eea5ae0

Adjust repetition penalty ..

ggerganov committed 2 years ago

Verified 93208cfb

Add link to recent podcast about whisper.cpp and llama.cpp

ggerganov committed 2 years ago

Verified 03ace14c

CI: CMake: Separate build and test steps (#376)

anzz1 committed 2 years ago

Verified e4412b45

Fix instruct mode broken by PR #354 (#409)

tjohnman committed 2 years ago

Verified f7dc43bc

Update issue template so people will use it (#404)

gjmulder committed 2 years ago

Verified ee8a7887

Older