Commits

Branch selector

 master

User selector

cebtenzzre

Datepicker

All time

Commit History

Commits on Dec 18, 2023

• decode : fix logits_valid for legacy API (ggerganov#4516)

  cebtenzzre

  authored
  2994f0c

Commits on Dec 17, 2023

• gguf-py : fail fast on nonsensical special token IDs (ggerganov#4489)

  cebtenzzre

  authored
  f7f468a

Commits on Dec 16, 2023

• llama : sanity checks for access to logits (ggerganov#4274)

  Show description for 8a5be3b
  

  cebtenzzre

  and

  ggerganov

  authored
  8a5be3b

Commits on Dec 13, 2023

• build : detect host compiler and cuda compiler separately (ggerganov#4414)

  cebtenzzre

  authored
  70f806b

Commits on Dec 12, 2023

• build : target Windows 8 for standard mingw-w64 (ggerganov#4405)

  Show description for 6138963
  cebtenzzre

  authored
  6138963

Commits on Dec 1, 2023

• build : enable libstdc++ assertions for debug builds (ggerganov#4275)

  cebtenzzre

  authored
  511f52c

Commits on Nov 30, 2023

• build : fix build info generation and cleanup Makefile (ggerganov#3920)

  Show description for 15f5d96
  cebtenzzre

  authored
  15f5d96

Commits on Nov 28, 2023

• ggml : restore abort() in GGML_ASSERT (ggerganov#4242)

  cebtenzzre

  authored
  64e64aa

Commits on Nov 27, 2023

• ggml : fix -Warray-bounds warning with gcc (ggerganov#4231)

  cebtenzzre

  authored
  f3b2698

Commits on Nov 15, 2023

• llama : restore prefix space in llama tokenizer (ggerganov#4081)

  cebtenzzre

  authored
  a6fc554

Commits on Nov 7, 2023

• gguf : track writer state, free unneeded tensors, cleanup (ggerganov#3871)

  cebtenzzre

  authored
  0a7c980

Commits on Nov 5, 2023

• cuda : fix disabling device with --tensor-split 1,0 (ggerganov#3951)

  Show description for 132d25b
  

  cebtenzzre

  and

  slaren

  authored
  132d25b

Commits on Nov 3, 2023

• llama : change yarn_ext_factor placeholder to -1 (ggerganov#3922)

  cebtenzzre

  authored
  3fdbe6b

Commits on Nov 2, 2023

• build : link against build info instead of compiling against it (ggerganov#3879)

  Show description for b12fa0d
  

  cebtenzzre

  and

  kassane

  authored
  b12fa0d

• cuda : fix RoPE after ggerganov#2268 (ggerganov#3897)

  cebtenzzre

  authored
  2fffa0d

Commits on Nov 1, 2023

• llama : fix llama_context_default_params after ggerganov#2268 (ggerganov#3893)

  cebtenzzre

  authored
  0eb332a

• llama : implement YaRN RoPE scaling (ggerganov#2268)

  Show description for 898aeca
  

  cebtenzzre

  and

  jquesnelle

  authored
  898aeca

Commits on Oct 29, 2023

• make : remove unnecessary dependency on build-info.h (ggerganov#3842)

  cebtenzzre

  authored
  2046eb4

Commits on Oct 27, 2023

• llama : correctly report GGUFv3 format (ggerganov#3818)

  cebtenzzre

  authored
  6d459cb

Commits on Oct 24, 2023

• server : add parameter -tb N, --threads-batch N (ggerganov#3584) (ggerganov#3768)

  Show description for ad93962
  

  3 peopleauthored
  ad93962

Commits on Oct 20, 2023

• convert : restore compat with old Falcon models (ggerganov#3680)

  cebtenzzre

  authored
  e78f3ef

Commits on Oct 15, 2023

• MPT : support GQA for replit-code-v1.5 (ggerganov#3627)

  cebtenzzre

  authored
  11bff29

Commits on Oct 7, 2023

• quantize : fail fast on write errors (ggerganov#3521)

  cebtenzzre

  authored
  f1782c6

Commits on Oct 5, 2023

• convert : update Falcon script for new HF config (ggerganov#3448)

  Show description for 48edda3
  

  cebtenzzre

  and

  logicchains

  authored
  48edda3

Commits on Oct 2, 2023

• convert : fix vocab size when not defined in hparams (ggerganov#3421)

  cebtenzzre

  authored
  1c84003

• cmake : increase minimum version for add_link_options (ggerganov#3444)

  cebtenzzre

  authored
  e78f0b0

• gguf : add BERT, MPT, and GPT-J arch info (ggerganov#3408)

  cebtenzzre

  authored
  29a404a

• gguf : general usability improvements (ggerganov#3409)

  cebtenzzre

  authored
  0fe3210

• cmake : make CUDA flags more similar to the Makefile (ggerganov#3420)

  Show description for 9476b01
  cebtenzzre

  authored
  9476b01

Commits on Sep 29, 2023

• llama : quantize up to 31% faster on Linux and Windows with mmap (ggerganov#3206)

  Show description for 2777a84
  

  cebtenzzre

  and

  ggerganov

  authored
  2777a84

Commits on Sep 28, 2023

• build : enable more non-default compiler warnings (ggerganov#3200)

  cebtenzzre

  authored
  bc39553

• gguf : basic type checking in gguf_get_* (ggerganov#3346)

  cebtenzzre

  authored
  2db94d9

• gguf : make token scores and types optional (ggerganov#3347)

  cebtenzzre

  authored
  ecf90b1

Commits on Sep 27, 2023

• gguf : fix a few general keys (ggerganov#3341)

  cebtenzzre

  authored
  20c7e1e

Commits on Sep 23, 2023

• examples : fix RoPE defaults to match PR ggerganov#3240 (ggerganov#3315)

  cebtenzzre

  authored
  51a7cf5

Pagination

PreviousNext