smallcloudai / refact

WebUI for Fine-Tuning and Self-hosting of Open-Source Large Language Models for Coding

https://refact.ai

smallcloudai/refact Issues

Model hosting improvements metabug
Closed 2 months ago2
LoRA metabug for v1.2
Closed 2 months ago2
Latest lora checkpoints for deepseek-coder/5.7b/mqa-base only generate 1 token to some requests
Updated 3 months ago1
[VS Code] Multiline completion not working in some cases
Updated 3 months ago
lora's "catastrophic forgetting" problem
Updated 3 months ago2
Finetune Problem
Updated 3 months ago12
Self host v1.4.0 MODEL always /infengine-v1/completions-wait-batch WAIT time out
Closed 3 months ago10
Support for HTTP proxies
Closed 4 months ago
Simple API key for OSS version, so people can expose docker port via reverse proxy
Closed 4 months ago1
Finetune of deepseek-coder fails
Closed 4 months ago7
Finetune improvement for better performance
Closed 4 months ago2
more files to process than processes
Closed 4 months ago4
Plugins should react faster to configuration changes in docker
Closed 4 months ago1
[ui] Finetune progress bar
Closed 4 months ago
Fix <th> number of columns in finetune
Closed 4 months ago
[CICL] cache flash_attn
Closed 4 months ago
Host Model for Embeddings for RAG
Closed 4 months ago
refact refuses to finetune if finds weird bytes in files
Closed 5 months ago2
docker image fails to start on mac m3
Updated 5 months ago8
Database not starting?
Closed 5 months ago2
Rotate server logs
Closed 5 months ago3
[metric] Store metrics in DB
Closed 5 months ago1
Finetune failed with "No train files provided"
Closed 5 months ago
GPU Filtering improvement
Closed 5 months ago
Self Hosted Chat Times Out VSCode
Updated 5 months ago
Add DeepSeek Coder models
Closed 6 months ago4
[finetune] Save and load loras in safetensors
Closed 6 months ago1
Add Code LLaMA
Closed 6 months ago4
add check for the minumum number of files for fine-tuning job
Closed 6 months ago
When Using Self-Hosted Server VS Code Plugin Does Not Send Requests
Closed 6 months ago2
Refactoring of the finetuning script
Closed 6 months ago1
[ui] Filtering error doesn't show
Closed 6 months ago3
[ui] Add a convenient link to the next page, after scanning is complete
Closed 6 months ago4
Without DB, web UI should still open, tell what the problem is
Updated 7 months ago
Could not scan repo with Refact/1.6B selected
Closed 7 months ago2
EPIC: Run Self-hosted version on CPU
Updated 7 months ago6
easy bug -- query_nvidia_smi doesn't have temp_celsius, ValueError: invalid literal for int() with base 10, "[N/A]"
Closed 7 months ago
[finetune] lora_r does not depend on the codebase size which contradicts Refact's official blog article
Closed 7 months ago1
Always "latest/best" now, instead of the clicked LoRA
Closed 7 months ago1
[finetune] Make starcoders finetune available for regular users
Closed 7 months ago2
[ui] Clicking LoRAs 2
Closed 7 months ago1
Handle OOM better on smaller/older GPUs, or bigger models on regular GPUs
Updated 7 months ago4
Multiple finetuned models metabug
Closed 8 months ago
EPIC: Add context-awareness on entire codebase
Updated 8 months ago
Review scratchpads for supported models
Closed 8 months ago
Improve contributing.md
Updated 8 months ago2
[backend] Model fails to load after deleting and re-adding it
Closed 8 months ago1
[ui] Cannot turn LoRA back on
Closed 8 months ago2
Assert when unloading Refact/1.6B/vllm model
Closed 8 months ago3
[finetune] Two git sources, only one works
Closed 8 months ago2