All tags
Model: "claude-2"
Adept Fuyu-Heavy: Multimodal model for Agents
fuyu-heavy fuyu-8b gemini-pro claude-2 gpt4v gemini-ultra deepseek-coder-33b yi-34b-200k goliath-120b mistral-7b-instruct-v0.2 mamba rwkv adept hugging-face deepseek mistral-ai nous-research multimodality visual-question-answering direct-preference-optimization benchmarking model-size-estimation quantization model-merging fine-tuning instruct-tuning rms-optimization heterogeneous-ai-architectures recurrent-llms contrastive-preference-optimization
Adept launched Fuyu-Heavy, a multimodal model focused on UI understanding and visual QA, outperforming Gemini Pro on the MMMU benchmark. The model uses DPO (Direct Preference Optimization), gaining attention as a leading tuning method. The size of Fuyu-Heavy is undisclosed but estimated between 20B-170B parameters, smaller than rumored frontier models like Claude 2, GPT4V, and Gemini Ultra. Meanwhile, Mamba was rejected at ICLR for quality concerns. In Discord discussions, DeepSeek Coder 33B was claimed to outperform GPT-4 in coding tasks, and deployment strategies for large models like Yi-34B-200K and Goliath-120B were explored. Quantization debates highlighted mixed views on Q8 and EXL2 quants. Fine-tuning and instruct-tuning of Mistral 7B Instruct v0.2 were discussed, alongside insights on RMS optimization and heterogeneous AI architectures combining Transformers and Selective SSM (Mamba). The potential of recurrent LLMs like RWKV and techniques like Contrastive Preference Optimization (CPO) were also noted.
1/2/2024: Smol tweaks to Smol Talk
claude-2 bard copilot meta-ai gemini-ultra chatgpt openai meta-ai-fair perplexity-ai prompt-engineering api json yaml markdown chatbot image-generation vpn browser-compatibility personality-tuning plugin-issues
OpenAI Discord discussions highlight a detailed comparison of AI search engines including Perplexity, Copilot, Bard, and Claude 2, with Bard and Claude 2 trailing behind. Meta AI chatbot by Meta is introduced, available on Instagram and Whatsapp, featuring image generation likened to a free GPT version. Users report multiple browser issues with ChatGPT, including persistent captchas when using VPNs and plugin malfunctions. Debates cover prompt engineering, API usage, and data formats like JSON, YAML, and Markdown. Discussions also touch on ChatGPT's personality tuning and model capability variations. "Meta AI includes an image generation feature, which he likened to a free version of GPT."