whats the term for when you think you have a cool idea and talk for ages to gemini and it only thinks for 9 secs
LLMS
-
Nvidia Parakeet TDT 0.6B V2 tops Hf ASR leaderboard
By
–
英伟达出了一个 ASR，叫 Parakeet TDT 0.6B V2，目前只支持英文，发布之后，直接冲到了 HF ASR Leaderboard 第一位。 亮点 自动加标点与大小写 精准的逐词时间戳（点击表格段落可播放） 支持最长 20 分钟音频转写（更长请看官方脚本） 强悍处理 数字口播 与 歌词识别 体验来看，感觉 ASR
-
Could Products Be Built Solely with LLM and Tools?
By
–
Could it have been built solely with llm+tools/mcp as the product? Everything I build with vibes gets overbuilt by default and finding myself thinking of how much to strip back to just llm plus tools. It’s worked perfectly for a couple of things I tried but ofc depends on if
-
Gemini 2.5 Pro Impresses Google Team with Special Features
By
–
It’s been so fun to see the Gemini team geeking out about this. Gemini 2.5 Pro is really something special!
-
Chain of Thought: How LLMs Think Recursively and Invoke Tools
By
–
Nope, it’s a thing called “chain of thought” where the LLM basically recursively invokes itself and outside tools to e.g. think up search terms, get results back, read them, not find it, back to the tank to think of new search terms, another tool invocation, etc.
-
Cerebras Powers 18x Speed Boost for Meta’s Llama API
By
–
Proud to have powered the 18x speed in Meta’s Llama API demo. We can't wait to see what you will build. @AIatMeta Cerebras Systems
-
Command A Achieves Top Bird Bench SQL Performance
By
–
Command A, our state-of-the-art generative model, is now the highest-scoring generalist LLM on the Bird Bench leaderboard for SQL! It outperforms other systems that rely on extensive scaffolding to tackle these SQL benchmarks, and instead delivers these results out-of-the-box,
-
Free Course: Pretraining LLMs with Upstage and DeepLearning.AI
By
–
Explore "Pretraining LLMs," a short course developed with @upstageai. DeepLearning.AI (@DeepLearningAI) Explore "Pretraining LLMs," a short course developed with @upstageai. While prompting or fine-tuning existing models is effective for many general language tasks, pretraining becomes essential for specialized domains or underrepresented languages in current models. This course guides you through the LLM pretraining pipeline—from data preparation to model evaluation—and introduces depth up-scaling, an innovative technique used by Upstage to train their Solar model family, cutting pretraining compute costs by up to 70%. Take the course for free: hubs.la/Q03kT_FP0 — https://nitter.net/DeepLearningAI/status/1918365268293607723#m
-
LlamaCon Highlights AI Innovation at Meta
By
–
LlamaCon got us seeing llamas everywhere — literally. 🦙 #LlamaCon @AIatMeta!
— SambaNova (@SambaNovaAI) 2 mai 2025LlamaCon got us seeing llamas everywhere — literally. #LlamaCon @AIatMeta
! -
Gemini Long Context Insights: Podcast Discussion with Experts
By
–
Have you been longing for more insight into how long context works in Gemini models? Check out this podcast discussion with @SavinovNikolay and @OfficialLoganK ⬇️ https://t.co/EkUsOWbOoI
— Jeff Dean (@JeffDean) 2 mai 2025Have you been longing for more insight into how long context works in Gemini models? Check out this podcast discussion with @SavinovNikolay and @OfficialLoganK