ML News

May 2023

• 25: BLIP-Diffusion: Image generation modelwith zero-shot editing, style transfer and more (tweet, paper)
• 24: QLoRA: 4-bit finetuning (tweet, paper, demo)
• 22: BigCode Evaluation harness for code LLMs (tweet)
• 18: VisualGLM-6B (code, model)
• 16: PEFT Whisper (tweet, code)
• 12: Spacy-llm, Integrating LLMs into structured NLP pipelines, Explosion (tweet, code)
• 11: ProteinGeneration: diffusion in prpotein sequence space (tweet, paper)
• 10: PaLM 2, Google (blog)
• 9: StarChat: Creating a Coding Assistant with StarCoder (tweet, blog, demo)
• 5: OpenLLaMA: An Open Reproduction of LLaMA (tweet, code, model)
• 4: StarCoder, A State-of-the-Art LLM for Code (tweet, blog, code, model)
• 3: Pi, Personal Intelligence (tweet, demo)

April 2023

• 28: StableVicuna, RLHF LLM Chatbot (tweet, blog)
• 28: FastChat-T5: Compact, commercial-friendly chatbot (tweet, code)
• 28: DeepFloyd IF: State of the art text-to-image model that can also generate text (text, demo)
• 25: HuggingChat (tweet, UI)
• 25: MOSS, a 16B tool-augmented language model (tweet, code)
• 25: NeMo Guardrails (NVIDIA), (blog, code)
• 25: Track Anything: Segment Anything Meets Videos (tweet)
• 24: Scaling Transformer to 1M tokens (tweet, code)
• 20: Run Whisper 70x faster with JAX and TPU (tweet)
• 19: StableLM (tweet)
• 16: CAMEL: Physics, Chemistry and Biology datasets (tweet)
• 15: Open Assistant (tweet, website)
• 15: ControlNet 1.1 (tweet)
• 7: SegGPT: Segmenting Everything in context (tweet)
• 6: Vicuna-7B weights are released (tweet)
• 6: StackLlama (tweet)
• 6: VideoCrafter: text to video model (tweet)
• 6: Generative Novel View synthesis (tweet)
• 5: SAM - Segment anything (tweet)
• 5: ChatArena, multi-agent game environments for LLMs (tweet)
• 5: Kandinsky 2.1 for image generation (tweet)
• 5: LLaMA-Adapter (tweet)
• 5: LatentVideo Diffusion Models for long video generation (tweet)
• 4: MolFeat, a hub of molecular featurizers (tweet)
• 4: LangChain announced their $10M seed round (tweet)
• 4: Kandinsky 2.1 (tweet)
• 4: IGEL, an instruction-uned German LLM (tweet)
• 4: Koala-13B: A Dialogue Model for Academic Research (tweet, blog)
• 4: Baize: An Open-Source chat model with PEFT (tweet)
• 3: Vicuna-13B weights are released (tweet)
• 3: A Survey of Large Language Models (tweet)

March 2023

• 30: BloombergGPT: A Large Language Model for Finance (tweet, paper)
• 30: Nucleotide Transformer, SOTA Genomics (tweet, code)
• 30: ColossalChat (blog, code)
• 29: GeoV-9b (tweet, code, weights, colab)
• 29: Spanish BERTIN GPT-J-6B Alpaca and Alpaca LoRA (tweet)
• 29: LLaMA Adapter (tweet, code, paper)
• 28: PRESTO dataset (github)
• 28: OpenFlamingo (tweet, blog)
• 28: Raven RWKV (RWKV finetuned on alpaca and codealpaca) (tweet, demo)
• 28: Cerebras-GPT (tweet, models)
• 28: GPT4All (tweet, code)
• 28: Replit Partners with Google Cloud (tweet)
• 27: LLaMA voice chat + Siri TTS (tweet)
• 26: Japanese Alpaca LoRA (tweet, demo, report)
• 26: LLaMA voice chat (tweet)
• 24: Text2Video-Zero (tweet, code)
• 24: Dolly (tweet, code, demo)
• 22: Alpaca LoRA as a chatbot (tweet, code).
• 20: Runway Gen-2 (tweet)
• 24: SwissBERT (tweet, blog)
• 17: Alpacoom: BLOOM fine-tuned on Alpaca's dataset using LoRA (tweet, model)
• 16: Alpaca LoRA: instruct tune LLAMA on consumer hardware (tweet, code)
• 14: Claude, Anthropic (blog)
• 14: ChatGLM-6B (code, model)