All tags

Model: "gpt-4"

    not much happened today
    not much happened today
    not much happened today
    not much happened to end the week
    Gemini (Experimental-1114) retakes #1 LLM rank with 1344 Elo
    s{imple|table|calable} Consistency Models
    a calm before the storm
    not much happened today
    AIPhone 16: the Visual Intelligence Phone
    Ideogram 2 + Berkeley Function Calling Leaderboard V2
    not much happened today
    a quiet weekend
    SciCode: HumanEval gets a STEM PhD upgrade
    RouteLLM: RIP Martian? (Plus: AINews Structured Summaries update)
    Mozilla's AI Second Act
    Hybrid SSM/Transformers > Pure SSMs/Pure Transformers
    The Last Hurrah of Stable Diffusion?
    Francois Chollet launches $1m ARC Prize
    HippoRAG: First, do know(ledge) Graph
    Qwen 2 beats Llama 3 (and we don't know how)
    Life after DPO (RewardBench)
    Cursor reaches >1000 tok/s finetuning Llama3-70b for fast file editing
    OpenAI's PR Campaign?
    Kolmogorov-Arnold Networks: MLP killers or just spicy MLPs?
    DeepSeek-V2 beats Mixtral 8x22B with >160 experts at HALF the cost
    $100k to predict LMSYS human preferences in a Kaggle contest
    Evals: The Next Generation
    LLMs-as-Juries
    FineWeb: 15T Tokens, 12 years of CommonCrawl (deduped and filtered, you're welcome)
    Multi-modal, Multi-Aspect, Multi-Form-Factor AI
    ReALM: Reference Resolution As Language Modeling
    DBRX: Best open model (just not most efficient)
    Andrew likes Agents
    World_sim.exe
    The world's first fully autonomous AI Engineer
    Fixing Gemma
    FSDP+QLoRA: the Answer to 70b-scale AI for desktop class GPUs
    Inflection-2.5 at 94% of GPT4, and Pi at 6m MAU
    Not much happened today
    Claude 3 just destroyed GPT 4 (see for yourself)
    Welcome Interconnects and OpenRouter
    Karpathy emerges from stealth?
    The Dissection of Smaug (72B)
    MetaVoice & RIP Bard
    RWKV "Eagle" v5: Your move, Mamba
    GPT4Turbo A/B Test: gpt-4-1106-preview
    RIP Latent Diffusion, Hello Hourglass Diffusion
    Sama says: GPT-5 soon
    1/10/2024: All the best papers for AI Engineers
    1/6-7/2024: LlaMA Pro - an alternative to PEFT/RAG??
    12/23/2023: NeurIPS Best Papers of 2023
    12/22/2023: Anyscale's Benchmark Criticisms
    12/21/2023: The State of AI (according to LangChain)
    12/20/2023: Project Obsidian - Multimodal Mistral 7B from Nous
    12/19/2023: Everybody Loves OpenRouter
    12/16/2023: ByteDance suspended by OpenAI
    12/15/2023: Mixtral-Instruct beats Gemini Pro (and matches GPT3.5)
    12/14/2023: $1e7 for Superalignment
    12/13/2023 SOLAR10.7B upstages Mistral7B?
    12/12/2023: Towards LangChain 0.1
    12/11/2023: Mixtral beats GPT3.5 and Llama2-70B
    12/10/2023: not much happened today
    12/8/2023 - Mamba v Mistral v Hyena
    12/7/2023: Anthropic says "skill issue"
    Is Google's Gemini... legit?