OS Week Highlights - Oct 9 - 15
Running209π’Note Nice tool to see and explore 3D generations
ostris/super-cereal-sdxl-lora
Text-to-Image β’ Updated β’ 390 β’ β’ 41
Runtime error33π₯CerealBoxMaker
Note Make your own cereal boxes with the super cereal lora above (and actually print them!)
stabilityai/japanese-stablelm-instruct-alpha-7b-v2
Text Generation β’ Updated β’ 292 β’ 22Note LLM in Japanese https://ja.stability.ai/blog/japanese-stablelm-instruct-alpha-7b-v2
HuggingFaceH4/zephyr-7b-alpha
Text Generation β’ Updated β’ 13.2k β’ β’ 1.1kNote Mistral trained with Direct Preference Optimization and with strong performance on MT Bench
thesephist/contra-bottleneck-t5-small-wikipedia
Text Generation β’ Updated β’ 616 β’ 11Note One among multiple BOttleneck T5 Models that allow editing text in latent space
zhiqings/dromedary-2-70b-qlora-delta-v0
Updated β’ 4Note Self-aligned LLM with minimal human supervision.
forestai/fireact_llama_2_7b
Text Generation β’ Updated β’ 14 β’ 14Note FreAct released llama and codellama fine-tuned using ReAct to use external tools
prometheus-eval/prometheus-7b-v1.0
Text2Text Generation β’ Updated β’ 157 β’ 30Note LLM trained to work as a GPT replacement to perform RLHF evaluations
Sleeping2π’Crystal
Note Introspective reasoning model to answer common sense questions
replit/replit-code-v1_5-3b
Text Generation β’ Updated β’ 22.9k β’ 289Note Replit very small code model
princeton-nlp/SWE-bench
Viewer β’ Updated β’ 21.5k β’ 22.4k β’ 86Note Benchmark from Princeton that tests if a model is able to solve GitHub's issues automatically. It contains over 2000 issue-PR pairs from 12 popular Python projects
OpenLemur/lemur-70b-v1
Text Generation β’ Updated β’ 1.23k β’ 45Note LLM optimized to work as backbone of agents
SkunkworksAI/BakLLaVA-1
Text Generation β’ Updated β’ 64 β’ 377Note Vision LLM based on Mistral 7B
Runtime error157π’InternLM XComposer
Note Vision LLM based on InternLM for textimage comprehension and composition. It can do text generation, image captioning, sporting, retrieval and selection.
alaa-lab/InstructCV
Image-to-Image β’ Updated β’ 103 β’ 9Note A bit old, but got lots of attention now: your tti models are vision generalists
fondant-ai/fondant-cc-25m
Viewer β’ Updated β’ 25.9M β’ 730 β’ 53Note Fondant contains 25 million image URLs with their CC License