Pinned
Maxime Rivest 🧙♂️🦙🐧
7,445 posts
Maxime Rivest 🧙♂️🦙🐧
@MaximeRivest
Searching for the wise way to interact with AI.
Experiments are on GitHub
Chat bot are likely not it.
DSPy is probably onto something, so I contribute to it.
Honey, if we buy 2 gpus (for 35 000$) we could probably save 200$/month in AI subscriptions, what do you think?
- Chrome DevTools is, by far, my favorite mcp server. By running these 2 commands in your terminal, you get to have Claude Code come into your browser with full power: Launch a chrome browser: > google-chrome --remote-debugging-port=9222
- This reminds me how Anthropic was the first to introduce: Projects Canvas Computer Use Coding CLI MCP and now Skills. We can say many things about Anthropic, but we cannot say that they don't actively innovate and share regarding LLM usage and ergonomics.Claude can now use Skills. Skills are packaged instructions that teach Claude your way of working.
00:00 - This morning, I thought I'd push Claude Code a little. I asked it to get speech-to-text running on my laptop using NVIDIA's 0.6b Parakeet model. Less than two hours later, I had speech-to-text better than OpenAI's Whisper running locally on my laptop CPU, transcribing as a
- Today, I pruned 87.24% of Qwen 30B for a sentiment classification task while keeping 100% of its accuracy. This means we get to use big models on gpus with not that much RAM (potentially running models that would normally require an H100 on 3090 type gpus)! Imagine pruning
- I started this morning, and it's now ready for you to try: > uv pip install maivi > maivi That will get you a free and open-source voice-to-text model running locally on your CPU/laptop—you don't need a GPU! The model is (according to benchmarks) better than Whisper, and ourThis morning, I thought I'd push Claude Code a little. I asked it to get speech-to-text running on my laptop using NVIDIA's 0.6b Parakeet model. Less than two hours later, I had speech-to-text better than OpenAI's Whisper running locally on my laptop CPU, transcribing as a
- Beautiful! You can now do: > uv pip install mcp2py dspy And in just 6 lines of Python code, you have an AI agent that can retrieve information through Google Chrome MCP DevTools. With the added bonus that you're only 1 or 2 steps away from doing prompt optimization for that
- Replying to @simonwWe used 70b models to create a synthetic dataset of about 50 000 examples and trained a bert based classifier on that. We then applied it to billions of records. With great success. Not specifically a llm finetune. We are moving into that soon. We are wondering if finetuning 1b
- It's hard to not like dspy with such results. Using dspy it seems that we can prompt gemma 1b to 'act' like sonnet 4 in a classification task. Until now we could only get that type of results with half-a-day of prompt engineering on 8b to 24b models depending on the subject being
- Without dspy and attachments I needed 3 scripts (each 500+ lines) AND I did not even know how good my llm parsing was..Now I just need 44 lines of code. Nothing more! 85% of the invoices perfectly correct .. almost there! 92% with optimised Gemini 2.5 pro
- MCP was, technically, entirely unnecessary. Psychologically, however, it is a masterpiece. If a company makes an MCP, you feel like you have permission to use AI with that API service. It stimulates you to think: "Is there an MCP for that app?" or to install an MCP for that
