Diego de las Casas
LightNews LightNews
banner
dlsq.bsky.social
Diego de las Casas
@dlsq.bsky.social
2.3K followers 300 following 21 posts
AI Scientist at Mistral AI.
Past: Google DeepMind.
🇧🇷 in 🇬🇧
Posts Replies Media Videos
Reposted by Diego de las Casas
tirelesstribal.bsky.social
Gotta wait until he double-crosses Indiana Jones to steal the Holy Grail, I'm afraid.
March 9, 2025 at 2:02 PM
dlsq.bsky.social
App store: apps.apple.com/us/app/le-ch...

Play store: play.google.com/store/apps/d...
‎Le Chat by Mistral AI
‎Le Chat combines powerful AI with extensive information on the web to help you rediscover the world. Enjoy natural conversations, real-time internet search, comprehensive document analysis, and much ...
apps.apple.com
February 6, 2025 at 6:08 PM
dlsq.bsky.social
check out more in our post:
mistral.ai/en/news/all-...
The all new le Chat: Your AI assistant for life and work | Mistral AI
Brand new features, iOS and Android apps, Pro, Team, and Enterprise tiers.
mistral.ai
February 6, 2025 at 6:08 PM
dlsq.bsky.social
Have you tried our 8B model?
huggingface.co/mistralai/Mi...
mistralai/Ministral-8B-Instruct-2410 · Hugging Face
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
huggingface.co
January 31, 2025 at 1:46 PM
dlsq.bsky.social
Mistral Small 3 is also available on many partner platforms:
- Ollama: ollama.com/library/mist...
- Kaggle: kaggle.com/models/mistr...
- Fireworks: fireworks.ai/models/firew...
- Together: together.ai/blog/mistral...

And many more soon!
mistral-small
Mistral Small 3 sets a new benchmark in the “small” Large Language Models category below 70B.
ollama.com
January 30, 2025 at 9:17 PM
dlsq.bsky.social
Performance of Mistral Small 3 Instruct model
huggingface.co/mistralai/Mi...
January 30, 2025 at 9:17 PM
dlsq.bsky.social
Mistral Small 3 Base model
huggingface.co/mistralai/Mi...
January 30, 2025 at 9:17 PM
dlsq.bsky.social
Mistral Small 3 architecture is optimised for latency while preserving high quality
January 30, 2025 at 9:17 PM
Reposted by Diego de las Casas
drmikewiser.bsky.social
I know, but it's just an application of one of my favorite memes:
Screen cap from one of the Thor movies featuring a dark haired pale skinned woman as Thor's sister Hela. She has her hand out stopping Thor's hammer (Mjölnir) in mid air. The hammer is labeled "It's basic biology". Hela is labeled "Advanced Biology"
January 21, 2025 at 7:07 PM
Reposted by Diego de las Casas
beenwrekt.bsky.social
In fact, statistical malpractice is the main driver of progress in machine learning. At some point, we need to come to terms with this.
November 22, 2024 at 2:40 PM
dlsq.bsky.social
Fsdp2 has a different policy for handling streams that is also worth a read
github.com/pytorch/pyto...
[RFC] Per-Parameter-Sharding FSDP · Issue #114299 · pytorch/pytorch
Per-Parameter-Sharding FSDP Motivation As we looked toward next-generation training, we found limitations in our existing FSDP, mainly from the flat parameter construct. To address these, we propos...
github.com
November 23, 2024 at 10:49 AM
dlsq.bsky.social
Pixtral Large:
- 123B decoder, 1B vision encoder, 128K sequence length
- Frontier multimodal model
- Maintains text performance of Mistral Large 2

HF weights: huggingface.co/mistralai/Pi...
Try it: chat.mistral.ai
Blog post: mistral.ai/news/pixtral...
Comparison table of various AI models across different benchmarks: Mathvista, MMMU, ChartQA, DocVQA, VQAv2, AI2D, and MM MT-Bench. Models are categorized into Open Weights, Closed, and Unreleased. Key models include Pixtral Large, Llama-3.2 90B, Gemini-1.5 Pro, GPT-4o, Claude-3.5 Sonnet, Llama-3.1 505B, and Grok-2. The table shows measured and reported performance scores, highlighting differences in model capabilities across various tasks. Pixtral Large excels in Mathvista, DocVQA, AI2D and MM MT-Bench benchmarks.
November 18, 2024 at 5:57 PM