[HN Gopher] BLOOMChat, a 176B parameter, Multi-lingual, fine tun...
       ___________________________________________________________________
        
       BLOOMChat, a 176B parameter, Multi-lingual, fine tuned chat
        
       Author : hatcherdogg
       Score  : 21 points
       Date   : 2023-05-19 19:44 UTC (3 hours ago)
        
 (HTM) web link (huggingface.co)
 (TXT) w3m dump (huggingface.co)
        
       | hatcherdogg wrote:
       | BLOOMChat is a 175B chat model able to have multilingual
       | conversations after being fine-tuned on English data. Built by
       | SambaNovaAI and Together by fine-tuning chat
        
       | Giorgi wrote:
       | it has some weird math problems, I asked to compare to ChatGPT
       | itself and it responded with that while ChatGPT is trained on 120
       | billion messages, bloomchat is trained on 1.7 billion messages,
       | thus bloomchat is trained on more data
       | 
       | When I asked which is more 1.7 or 120, it said 1.7 is greater
       | number and then started spewing complete garbage math how 1.7 -
       | 120 = 60 and since 60 is more than 0 then 1.7 is more than 120.
       | 
       | Utter garbage
        
         | jug wrote:
         | Try to not use it on math. Only GPT-4 has reasonable
         | performance there. GPT 3.5 is also pretty awful. It's
         | apparently extremely hard for any LLM to actually understand
         | math. Maybe because they're language models, not math models,
         | so math is a pretty far fetched "emergent property".
        
         | jxy wrote:
         | Bloom models are hopelessly under-trained. This one is worse
         | than a 13B Vicuna.
        
         | Semaphor wrote:
         | I've got similar garbage out of ChatGPT (though tbf, pre-GPT4),
         | I don't think LLMs can _understand_ math
        
       ___________________________________________________________________
       (page generated 2023-05-19 23:02 UTC)