cnvrs

🚧 🚧

[cnvrs @cnvrsai]

@nick_plaid https://t.co/ZlUD1Q9cg3

[cnvrs @cnvrsai]

@nick_plaid of course! check out the link in my bio

[Agent Sid  @PeoplesSpirit]

@simonw Hi Simon, check this @cnvrsai app if you have a iPhone Pro model, you shouldn’t have any issues for 4bit quantised 4B models:

[britt @p3ery]

in the latest version of @cnvrsai, dramatically improved coherence for long prompts & long conversations!

see this example below of before & after with OLMoE 1B-7B Instruct — same model, same prompt, same paper pdf attached — night and day response quality!

[britt @p3ery]

in the next version of @cnvrsai…

[britt @p3ery]

now chat with online models from chatgpt or o3-mini, to deepseek r1, or any other chat-completions compatible provider!

or chat with private, offline models as always

…in the next version of @cnvrsai

[britt @p3ery]

now run the deepseek r1 family of models, customize your default character, & try the less-sassy “Assistant” in the character gallery

…all in the latest version of @cnvrsai

[britt @p3ery]

now control context & sampling params, and see model performance metrics — in the latest version of @cnvrsai!

[cnvrs @cnvrsai]

@p3ery @caio_carduz @Thom_Wolf llama 3.2 1b & 3b run great! 🏎️💨

[britt @p3ery]

gemma 2 2b is incredible!

it’s fast, cogent, pleasant to chat with, handles multiturn conversations with aplomb

& surprisingly smart given its size!

check out how it cruises thru implementing fibonacci in three different languages — rust, typescript & swift

in @cnvrsai

[britt @p3ery]

@atbeme @yacineMTB just saying, you totally *could* run llama on your phone..!

[cnvrs @cnvrsai]

@jormungandr1980 thanks for your interest! a fresh testflight build is on the way

[big snooood @bitte_digga]

@cnvrsai this is amazing considering organic chemistry is a complex topic and I only gave it 51 pages to work with.

[britt @p3ery]

Hermes 2 Pro Llama 3 longs for internet access 🥲

[britt @p3ery]

phi 3 mini is fast, smart, and best of all, runs like butter on your phone with the latest @cnvrsai!

[cnvrs @cnvrsai]

get your ggufs here:

https://t.co/qG0ro7gyl7

https://t.co/il6kGKCNr6

[cnvrs @cnvrsai]

Try @google’s new Gemma model for yourself!

[britt @p3ery]

@ggerganov @Google runs beautifully on @cnvrsai 😍

[britt @p3ery]

@WuMinghao_nlp so speedy! ⚡️

[britt @p3ery]

@Euclaise_ so exciting!!

@h2oai’s new danube 1.8b chat is quite good as well!

tiny models 🤝

[britt @p3ery]

the new 1.6B StableLM 2 from @StabilityAI is wickedly fast!

run it and hundreds of other models from @huggingface in @cnvrsai