Why do speech AI models still lag behind text AI? Researchers at CUHK and Huawei propose TextPro-SLM — an approach that shrinks the gap by making spoken input look more like text input. Instead of tweaking the output, they redesign the input side with a unified speech
LLMS
-
Subscription of $200 and loop without token spending in training
By
–
hehe luckily it has been with subscription ($200) and in a loop where there is a lot of execution without token spending during model training.
-
Learning Path for LLM Serving Engines: vLLM, SGLang, TensorRT-LLM
By
–
How to go about learning all of this? 1st: Start with the serving engine view – vLLM: PagedAttention, continuous batching, prefix caching, CUDA graphs – SGLang: RadixAttention/prefix reuse, speculative decoding, MoE, structured/agent workloads – TensorRT-LLM: NVIDIA peak
-
Unifying Codex, agents, and chat boosts value for users and valuation
By
–
Folding Codex, agents, and chat into one app makes sense for users and even more sense for the valuation.
-
Should you really say please and thank you to AI?
By
–
Should You Really Say Please And Thank You To AI? Millions of people now say please and thank you to #AI #chatbots, but does #politeness actually improve the answers or simply waste time and energy? As AI becomes more human-like, the bigger question may be how our #conversations
-
Request for long context benchmarks for decode and prefill
By
–
Give me long context benches for decode and prefill
-
Command /usage shows breakdown of skills, MCPs, plugins
By
–
Run /usage to see a breakdown of specific skills, MCPs, and plugins that use your tokens.
-
AI Week Recap: Model Drops, Memory Overhaul, IPO, and Federal Bill
By
–
The past week in AI was genuinely wild. Model launches. A $965B IPO filing. Memory that rewrites itself. And a federal AI bill nobody saw coming. Here's your May 31–June 5 recap What we're covering:
→ 6 major model drops
→ ChatGPT memory overhaul
→ Google's new personal -
Lab will take care of small specialized models as future
By
–
My lab will take care of this I had a stream in October 2024 saying small and specialized models are the future that I need to find