SOTA Speech and Text Translation

Hello folks, here’s what we have today;

Our picks

1/

SeamlessM4T - The first, all-in-one, multimodal translation model from Meta. The model supports nearly 100 languages for input (speech + text), 100 languages for text output and 35 languages (plus English) for speech output. It outperforms Whisper on the benchmarks and is available for download on Github.

2/

GPT-3.5 Turbo fine-tuning and API updates. Open AI has finally made fine-tuning with GPT 3.5 Turbo available, with the GPT-4 fine-tuning coming in a few months. Plus, OpenAI is changing the API endpoint for old GPT-3 models with the previous endpoint deprecating on January 4th, 2024.

3/

IDEFICS - An open-access reproduction of Flamingo. Flamingo is the SOTA vision-language model from Deepmind. The team at Hugging Face has made a replication of it using Llama v1 and CLIP. IDEFICS is comparable in performance with the original closed-source Flamingo model across various image-text understanding benchmarks.

4/

Visualising AI - A unique project from Deepmind with 13 artists coming together to create artworks about AI.

from our sponsor

Take a demo, get a $150 Nike gift card.

Running out of patience for manual expense reports? Divvy can help you speed through expense reporting (like Nikes can help you speed through your workout). Take a demo to learn more, and we’ll send you a $150 Nike gift card.*

In just 20 minutes, you’ll discover how to:

• Automate expense reports

• Take control of your budget and spending

• Access scalable business credit to grow your business

Claim your offer
*Terms and conditions apply: see offer page for more details

 From the community 

• AI startups - Sell work, not software.

 Cool Tools  trending product launches from the last 24 hours

• Accountabilabuddy by Summit - Your AI buddy that sends a simple text to motivate and hold you accountable.

• Gamma 1.0 - AI text, visuals, and now images together in one tool for presentations.

• AI office - Starter kit for building your own version of AI town.

• GuestLab - Hours of guest research delivered in seconds.

• Homation - Easy way to create a smart home with constructor.

• Rainbow AI - Precise precipitation and rain forecasting.

• Laxis - Repurpose audio into engaging content with a single click.

• SEC Insights AI - Revolutionizing SEC document analysis.

• Ollama - The easiest way to run LLMs locally.

• ElevenLabs - The voice-generating tools launch out of beta.

 From the network 

• How to use AI to save time & grow your business.

• Cursor - The AI-first code editor with Aman Sanger of Anysphere.

• Request for startups from Weekend Fund - Synthetic Humans.

 Ben’s Bites News  top posts from the last 24 hours

• Google and YouTube are trying to have it both ways with AI and copyright.

• Teaching robots novel actions through natural language input with reward functions.

• Google co-founder Sergey Brin on leaving retirement to work on AI.

• Mendaera raises a $24M Series A to develop a collaborative robotic AI system for healthcare providers.

• IBM taps AI to translate COBOL code to Java.

• Tiger Global is selling a 2.1% stake in AI startup Cohere at a $3B valuation, up 40% from June 2023. Tiger will retain a ~5% stake.

• Nvidia and VMware extend partnership to help companies iterate on open models.

• AI company Hypergiant Industries snapped up by PE firm Trive Capital.

• Meta confirms AI ‘off-switch’ incoming to Facebook, Instagram in Europe.

• Portkey secures $3M funding to create AI apps in 2 minutes.

Ben’s Bites Insights

We have 2 databases that are updated daily which you can access by sharing Ben’s Bites using the link below;

• All 10k+ links we’ve covered, easily filterable (1 referral)

• 6k+ AI company funding rounds from Jan 2022, including investors, amounts, stage etc (3 referrals)