News Roundup 54
Created on March 7|Last edited on March 7
Comment
QwQ-32B pushes RL-driven reasoning with fewer parameters
QwQ-32B achieves DeepSeek R1-level performance with 20x fewer parameters using reinforcement learning.
AMD introduces Instella, a fully open 3B parameter language model
Instella, trained on AMD GPUs, outperforms similar open models and rivals top open-weight models.
CoreWeave targets $35B valuation in IPO
AI cloud firm CoreWeave files for Nasdaq IPO, aiming to raise $3 billion amid rapid growth.
Evaluating Claude 3.7 Sonnet: performance, reasoning, and cost optimization
Claude 3.7 Sonnet introduces extended reasoning, improved coding, and multimodal capabilities with a 200K token context.
Add a comment