Search code, repositories, users, issues, pull requests...

Refactoring

Fix(server): stopped_word always true

Trivial changes to code that most beginner devs (or those who want a break) can tackle. e.g. UI fix

#7204 opened May 10, 2024 by slaren

Loading…

bugfix

fixes an issue or bug

Add support for properly optimized Windows ARM64 builds with LLVM and MSVC

Trivial changes to code that most beginner devs (or those who want a break) can tackle. e.g. UI fix

server/webui

#7198 opened May 10, 2024 by kurikomoe

Loading…

devops

improvements to build systems and github actions

CUDA: add FP32 FlashAttention vector kernel

Generally require indepth knowledge of LLMs

#7191 opened May 10, 2024 by max-krasnyansky

Loading…

Nvidia GPU

Issues specific to Nvidia GPUs

performance

Speed related topics

Generally require indepth knowledge of LLMs

#7188 opened May 9, 2024 by JohannesGaessler

Loading…

main : add token healing

New feature or request

convert-hf : support bfloat16 conversion

Generally require indepth knowledge of LLMs

#7187 opened May 9, 2024 by mare5x

Loading…

performance

Speed related topics

ggml : rewrite silu and softmax for cpu

Generally require more time to grok but manageable by beginner to medium expertise level

#7158 opened May 9, 2024 by compilade

Loading…

3 of 4 tasks

Refactoring

Server: Fix system_prompt handling

Generally require indepth knowledge of LLMs

#7154 opened May 9, 2024 by jart

Loading…

bugfix

fixes an issue or bug

help wanted

Extra attention is needed

perplexity: add BF16 vs. FP16 results

Trivial changes to code that most beginner devs (or those who want a break) can tackle. e.g. UI fix

#7153 opened May 8, 2024 by ngxson

Loading…

documentation

Improvements or additions to documentation

need feedback

Testing and feedback with results are needed

chore: Add model vocab support

Trivial changes to code that most beginner devs (or those who want a break) can tackle. e.g. UI fix

#7150 opened May 8, 2024 by JohannesGaessler

Loading…

New feature or request

tokenization: no double BOS tokens

Generally require more time to grok but manageable by beginner to medium expertise level

#7117 opened May 7, 2024 by teleprint-me • Draft

Refactoring

Scripting & documenting debugging one test without anything else in the loop.

Generally require more time to grok but manageable by beginner to medium expertise level

#7107 opened May 6, 2024 by JohannesGaessler

Loading…

documentation

Improvements or additions to documentation

Add left recursion check: quit early instead of going into an infinite loop

Trivial changes to code that most beginner devs (or those who want a break) can tackle. e.g. UI fix

#7096 opened May 6, 2024 by josh-ramer

Loading…

bugfix

fixes an issue or bug

Script to convert Grok-1 weights from raw JAX pickle files. enhancement

Generally require more time to grok but manageable by beginner to medium expertise level

#7083 opened May 5, 2024 by nuchi

Loading…

New feature or request

python

python script changes

docs: Fix typo and update description for --embeddings flag

Generally require more time to grok but manageable by beginner to medium expertise level

#7058 opened May 3, 2024 by heiner • Draft

documentation

Improvements or additions to documentation

Added support for the ArcticForCausalLM.

Trivial changes to code that most beginner devs (or those who want a break) can tackle. e.g. UI fix

#7026 opened May 1, 2024 by louixs

Loading…

New feature or request

Fix flash attention for ROCm

Generally require more time to grok but manageable by beginner to medium expertise level

#7020 opened May 1, 2024 by fairydreaming

Loading…

New feature or request

add chatglm3-6b model support [help wanted]

Generally require indepth knowledge of LLMs

#7011 opened Apr 30, 2024 by jdecourval • Draft

New feature or request

help wanted

Extra attention is needed

new tokenizer-verifier tool to check gguf tokenizer parameters

Generally require more time to grok but manageable by beginner to medium expertise level

#6999 opened Apr 30, 2024 by mnlife • Draft

New feature or request

server: avoid breaking KV cache when prompt >= n_ctx

Trivial changes to code that most beginner devs (or those who want a break) can tackle. e.g. UI fix

#6988 opened Apr 29, 2024 by anisse

Loading…

New feature or request

move ndk code to a new library

Generally require more time to grok but manageable by beginner to medium expertise level

#6958 opened Apr 28, 2024 by prfd • Draft

android

Issues specific to Android

Refactoring

Option to split during conversion

Trivial changes to code that most beginner devs (or those who want a break) can tackle. e.g. UI fix

#6951 opened Apr 27, 2024 by eltonkola

Loading…

New feature or request

help wanted

Extra attention is needed

python

python script changes

Updated server_queue to delete tasks from queue when server is shutdown. Feature Request #6421

Generally require more time to grok but manageable by beginner to medium expertise level

#6942 opened Apr 27, 2024 by christianazinn • Draft

demo

Demonstrate some concept or idea, not intended to be merged

Implemented basic interface for llamacheck and link to weights, adapt…

Trivial changes to code that most beginner devs (or those who want a break) can tackle. e.g. UI fix

#6941 opened Apr 27, 2024 by rahsuri

Loading…

demo

Demonstrate some concept or idea, not intended to be merged

Fix clip build on windows + clang

Generally require more time to grok but manageable by beginner to medium expertise level

#6940 opened Apr 27, 2024 by Ferruolo

Loading…

bugfix

fixes an issue or bug