LLaMa Now Goes Faster on CPUs

Incredible sorcery.

LLaMA Now Goes Faster on CPUs

Code Generation on HumanEval

Top two Agents for Code Generation use Testing or Debugging at their core.

We empirically show that LDB significantly improves code generation accuracy and achieves state-of-the-art performance in program debugging, by segmenting the programs into basic blocks and tracking the intermediate values.

Link

Agent-based technology, 2003

Agent Technology: Enabling Next Generation Computing

SWE Agent Prompts

Link

Octopus v2: On-device language model for super agent

“When it comes to function calling, employing RAG-based (Lewis et al. [2020], Mao et al. [2020], Li et al. [2022], Jiang et al. [2023]) or context-augmented (Ram et al. [2023]) methods requires processing about 1000 tokens for each call, resulting in costs of approximately 0.01 USD. In practical applications, where hundreds of function calls may be made, the cumulative cost can be much. Additionally, the potential for privacy violations deters many from using GPT-4, amid concerns that sensitive information might be exposed."

Link

Banger

Design System Consistency

Link

Artificial Analysis

AI Benchmarks

Cloudflare Acquisitions

Cloudflare acquires PartyKit to allow developers to build real-time multi-user applications

Cloudflare acquires Baselime to expand serverless application observability capabilities

Seems CloudFlare is all-in on Sunil's vision of stateful serverless