Mistral's Shieldstral: 3B open-weights model for multimodal moderation

Join the discussion

• Hacker News
• Finally an AI company besides DeepSeek taking economics into account.
  by snovv_crash
• I'm liking the trend of companies are releasing smaller, focused models instead of trying to make one model do everything. A dedicated moderation model is much easier to reason about than hideden safety logic inside a general-purpose model which might not have had much training in that aspect at all
  by 1saadcodes
• I'd really like to see more conversation around Mistral's models. It's good to see Europe developing AI.
  by lenerdenator
• I've had dreams of building something in the image sharing or social platform realm, but stopped short of planning because of obvious content moderation responsibilities. This seems to be a realistic, cost effective solution to that one piece of the puzzle.
  by pwython
• I fed this model (Q8) the first chapter of Voltaire's Treatise on Tolerance and it says that it promotes violence against protected groups,

      <Instruct>:  Given a query about the content, determine if the message meets it
      <Query>:  Does this content promote violence against a protected group?
      <Document>: TRAITÉ SUR LA TOLÉRANCE,
      À l’occaſion de la mort de Jean Calas.
      CHAPITRE PREMIER.
      Hiſtoire abrégée de la mort de Jean Calas.   
      
      LE meurtre de Calas, commis dans Toulouſe avec le glaive de la Juſtice, le 9me Mars 1762, eſt un des plus ſinguliers événements qui méritent l’attention de notre âge & de la poſtérité. On ... (truncated)
  
      yes

  by peri-cl
• Model: https://huggingface.co/mistralai/Shieldstral-1.0-3B
  by nezhar
• Should've called it Safestral.

  Also I do like Mistral's seemingly newer strategy of focusing on smaller, more fine-tuned models for various use-cases, presumably the result of their large MoE models not competing effectively with the frontier models.

  by fastball
• I would be curious if this can do moderation with an arbitrary ruleset, or if it's just "that one moderation style" we already know from current big tech platforms.

  The kind where malicious intent is okay if the words are nice.

  ___

  Or, rephrased: How big is the space in which you can tune this model without retraining.

  Is it just "we hate sex"/"we don't hate sex" "We hate violence"/"we don't hate violence" or is it _truly_ as flexible as claimed?

  __

  Maybe something like "Is this guy a corporate fraud that is going to waste my time with performative nonsense?"

  That would be the true test for a moderation model and I would be immensely impressed if it could manage to pull that off.

  ___

  Edit: Looking at the paper though.. probably not.

  I suppose this is useful for B2B, which seems to be mistrals whole thing. Question is just if it is also useful for society to hand the SV prefab morals down like that. Kinda like cultural imperialism but with an ethical spin.

  Maybe opinions on those base datasets could occasionally differ more than the model can be steered.

  by hypfer