usages run out so fast before you even notice and it always keeps switching to grok
specially now since claude code usage lasts much longer than before,
specially now since claude code usage lasts much longer than before,
Also on a tangent grok sucks now , specially for development tasks
you can't rely on it for longer tasks, it literally starts hallucinating and mixing up varaibles it created a couple of hours before. literally ruins working features
and don't get me started on the caching, it is so inefficient it eats up your usage like crazy
you can't rely on it for longer tasks, it literally starts hallucinating and mixing up varaibles it created a couple of hours before. literally ruins working features
and don't get me started on the caching, it is so inefficient it eats up your usage like crazy
Forwarded from Addis AI
We open-sourced Addis Scribe Streaming, realtime speech recognition for Amharic.
It beats Google Chirp 3 on both accuracy and speed:
Word error rate: 28.4% vs 32.5%
First words on screen: 1.56s vs 6.67s
Final text after you stop talking: 30ms vs 1,411ms
On natural speech (WAXAL) we beat Google Chirp 3 in both streaming and batch mode. On FLEURS we went from 35.0% to 19.8% WER.
Hohe remains the most accurate offline Amharic model in our WAXAL test, and on FLEURS the two are within a point of each other (19.8% vs 20.6%). We built Addis Scribe for the live and edge inference side, where text has to appear while people are still talking.
Trained on 5,000+ hours of Amharic speech. Test sets were never used in training. Per-clip results and scripts are in the repo.
Try it😊 : https://docs.addisassistant.com/docs/playground/speech-to-text
API and SDK docs: docs.addisassistant.com
HuggingFace🤗: huggingface.co/addisai/addis-scribe-streaming
@addisassistantai
It beats Google Chirp 3 on both accuracy and speed:
Word error rate: 28.4% vs 32.5%
First words on screen: 1.56s vs 6.67s
Final text after you stop talking: 30ms vs 1,411ms
On natural speech (WAXAL) we beat Google Chirp 3 in both streaming and batch mode. On FLEURS we went from 35.0% to 19.8% WER.
Hohe remains the most accurate offline Amharic model in our WAXAL test, and on FLEURS the two are within a point of each other (19.8% vs 20.6%). We built Addis Scribe for the live and edge inference side, where text has to appear while people are still talking.
Trained on 5,000+ hours of Amharic speech. Test sets were never used in training. Per-clip results and scripts are in the repo.
Try it
API and SDK docs: docs.addisassistant.com
HuggingFace🤗: huggingface.co/addisai/addis-scribe-streaming
@addisassistantai
Please open Telegram to view this post
VIEW IN TELEGRAM
Please open Telegram to view this post
VIEW IN TELEGRAM
Forwarded from Ge'ez Tech® ግዕዝ ቴክ
አዲሱ የመንግስት አዋጅ በተለያዩ ዘርፎች የሳይበር ደህንነት ጥበቃን ያጠናክራል፣ እና ብቁ የCybersecurity ባለሙያዎችን የማሰማራት ግዴታንም ያካትታል።
https://academy.geezsecurity.com/course/gtst
#GeezTech #GTSTv2 #CyberSecurity #EthicalHacking #CyberSecurityJobs #Ethiopia
Please open Telegram to view this post
VIEW IN TELEGRAM
Been using opus 5.5 on medium for almost 3 hours and didn't even hit 50% of my 5 hour limit
the value is crazy
the value is crazy
Amir A
Been using opus 5.5 on medium for almost 3 hours and didn't even hit 50% of my 5 hour limit the value is crazy
Anthropic really did turn things around a few months ago. They were known as the expensive model with terrible limits. I still remember the days when I hit my 5-hour limit before the first hour even ended.