[HN Gopher] Qwen (Tong Yi Qian Wen ) chat and pretrained large l...
       ___________________________________________________________________
        
       Qwen (Tong Yi Qian Wen ) chat and pretrained large language model
       by Alibaba Cloud
        
       Author : slyall
       Score  : 17 points
       Date   : 2023-09-27 21:33 UTC (1 hours ago)
        
 (HTM) web link (github.com)
 (TXT) w3m dump (github.com)
        
       | behnamoh wrote:
       | Didn't the chinese also have a model with >1 trillion parameters
       | when GPT-3 came out? Didn't the chinese also tweak the lmsys
       | leaderboard to make their model appear on top? It's sad but
       | unfortunately as with so many other things coming from china, one
       | has to take their claims with a pinch of salt.
        
         | booleandilemma wrote:
         | It's funny and heartening to me that despite their larger
         | population size, we still out-innovate them.
        
       | b1n wrote:
       | What happens when you ask about Tiananmen Square?
       | 
       | (Asking for redwood, who's comment was inexplicably flagged).
        
       | version_five wrote:
       | I briefly played with this a couple months ago and it looked
       | promising. Have there been more developments?
        
         | euclaise wrote:
         | There's a new 7B version that was trained on more tokens, with
         | longer context, and there's now a 14B version that competes
         | with Llama 34B in some benchmarks.
        
       | Havoc wrote:
       | Looks like the mistral model beats this slightly at 7b. Though
       | this has a 14b and a chat tuned one so may be better for some
       | uses.
       | 
       | Qwen models had compatibility issues last time I tried it though.
        
       | redwood wrote:
       | What happens when you ask about Tiananmen Square and 1989?
        
         | euclaise wrote:
         | https://www.reddit.com/r/LocalLLaMA/comments/16sw4na/qwen_is...
        
           | hashtag-til wrote:
           | Why is this flagged? Honest question.
        
             | dnd_was_evil wrote:
             | Because flagging in hn is broken and used to hide ideas
             | people don't like.
             | 
             | Usually it's attacks on the California Ideology but there
             | are enough stooges sympathetic to Chinese authoritarianism
             | that attacks on China are also removed.
        
           | class4behavior wrote:
           | I get that someone would do it anyway, but why would the
           | poster want to be the one helping an authoritarian entity fix
           | these loopholes? smh
        
         | dnd_was_evil wrote:
         | [flagged]
        
       ___________________________________________________________________
       (page generated 2023-09-27 23:00 UTC)