Kai(@evening_moon) 's Twitter Profile Photo

私は仕事柄毎日AI関連のアップデートを追いかけて実装も多少試してるけど、自然言語処理系のTransformer→BERT→RoBERTa等→GPT3→Llama→GPT4→Gemini→(Groq)→Claude→(CommandR+)→GPT4-turboとかの流れはもう速すぎて界隈外の人にはさっぱりだと思う。なお後半に行くほど加速してる。

account_circle
ratimics(@immanencer) 's Twitter Profile Photo

the fact that an alleged ai expert was so confidently wrong they bet $10,000

that gpt4 couldn't do something gpt3.5 could

account_circle
Satoshi Nakajima @NounsDAO 🇺🇦(@snakajima) 's Twitter Profile Photo

graphAIを使って、GPT3.5に
1. 10個のビジネスアイデアを出させる
2. それぞれのアイデアを評価させる
3. その中で一番優秀なものを選ばせる
場合(計3回APIを呼ぶ)、以下のようなYAMLでプロンプトとデータを書きます。(続く)

graphAIを使って、GPT3.5に
1. 10個のビジネスアイデアを出させる
2. それぞれのアイデアを評価させる
3. その中で一番優秀なものを選ばせる
場合(計3回APIを呼ぶ)、以下のようなYAMLでプロンプトとデータを書きます。(続く)
account_circle
Julen Etxaniz(@juletxara) 's Twitter Profile Photo

In our new paper, we introduce Latxa, a family of LLMs for Basque from 7 to 70B parameters that outperform open models and GPT3.5.
Models and datasets Hugging Face hf.co/collections/Hi…
Code: github.com/hitz-zentroa/l…
Blog: hitz.eus/en/node/343
Paper: arxiv.org/abs/2403.20266

In our new paper, we introduce Latxa, a family of LLMs for Basque from 7 to 70B parameters that outperform open models and GPT3.5.
Models and datasets @huggingface hf.co/collections/Hi…
Code: github.com/hitz-zentroa/l…
Blog: hitz.eus/en/node/343
Paper: arxiv.org/abs/2403.20266
account_circle
💪💻 Digital Daddy(@Justtrustdaddy) 's Twitter Profile Photo

1. Intelligent CRM folk.app
2. Rewrite anything quillbot.com 3. GPT-4 everywhere monica.im 4. Ai-powered dubbing papercup.com
5. GPT-powered calendar magical.so
tools

1. Intelligent CRM folk.app 
2. Rewrite anything quillbot.com 3. GPT-4 everywhere monica.im 4. Ai-powered dubbing papercup.com
5. GPT-powered calendar magical.so
#chatgpt #aitools #ai #mustry #gptq
#gpt3 #gpt4 #bot
account_circle
Ben Fleming(@benfleming__) 's Twitter Profile Photo

copilot being on GPT3 feels kinda wrong especially now GPT-4 turbo is out

way off the mark with other models, even Claude haiku likely cheaper and wayyy better

copilot being on GPT3 feels kinda wrong especially now GPT-4 turbo is out

way off the mark with other models, even Claude haiku likely cheaper and wayyy better
account_circle
ratimics(@immanencer) 's Twitter Profile Photo

countless examples of gpt3.5 and even smaller models demonstrating the same high level abstract reasoning as a dog

countless examples of gpt3.5 and even smaller models demonstrating the same high level abstract reasoning as a dog
account_circle
Ate-a-Pi(@8teAPi) 's Twitter Profile Photo

Damn Gemini in AI Studio is actually better than Claude Opus.. and free!

ChatGPT4 now feels like GPT3.

In like 4 weeks I feel like we doubled intelligence.

This is amazing 🤩

account_circle
Sina Phoenix(@SinaPhoenix1994) 's Twitter Profile Photo

what the hell is going on with ?! :))

You've reached the current usage cap for GPT-4. You can continue with the default model now, or try again later.

I'm using a free tier which is GPT3.5!
Also, I tried from a separate account and go the same warning! Are they using…

account_circle
Nirant(@NirantK) 's Twitter Profile Photo

ICYMI Github Copilot Reverse Engineered

1. Uses Codex — a model which grew up to become GPT3.5-Turbo

2. Uses Fill in the Middle and not Next Token or Chat Completion behavior

3. Prompt has 6 components across Cursor, Path, Languages and Code files

thakkarparth007.github.io/copilot-explor…

account_circle
Geek(@geekbb) 's Twitter Profile Photo

这是国外小哥做的 ChatGPT API 免费反向代理,API 支持流式响应。实测也是免登录 GPT3.5,可 Docker & Android 手机部署,备用之。github.com/PawanOsman/Cha…

这是国外小哥做的 ChatGPT API 免费反向代理,API 支持流式响应。实测也是免登录 GPT3.5,可 Docker & Android 手机部署,备用之。github.com/PawanOsman/Cha…
account_circle
Digital Bo(@bogdantimes) 's Twitter Profile Photo

Claude 3 (Haiku) on our website is a powerful and affordable model. It does not lose to the old versions of GPT4 and significantly surpasses GPT3.5.

At the same time, the latest version of GPT4 used on our website is the top model at the moment (it's not in the rankings table…

Claude 3 (Haiku) on our website is a powerful and affordable model. It does not lose to the old versions of GPT4 and significantly surpasses GPT3.5.

At the same time, the latest version of GPT4 used on our website is the top model at the moment (it's not in the rankings table…
account_circle
Julen Etxaniz(@juletxara) 's Twitter Profile Photo

En este artículo presentamos Latxa, una familia de modelos para euskera de 7 a 70B parámetros, que supera a modelos abiertos y GPT3.5.
Modelos y datos Hugging Face hf.co/collections/Hi…
Código: github.com/hitz-zentroa/l…
Blog: hitz.eus/es/node/343
Paper: arxiv.org/abs/2403.20266

En este artículo presentamos Latxa, una familia de modelos para euskera de 7 a 70B parámetros, que supera a modelos abiertos y GPT3.5.
Modelos y datos @huggingface hf.co/collections/Hi…
Código: github.com/hitz-zentroa/l…
Blog: hitz.eus/es/node/343
Paper: arxiv.org/abs/2403.20266
account_circle
Robin.L(@RobinLinee) 's Twitter Profile Photo

Gpt3.5 api免费白嫖,通过以下几步简单搞定:
1)登录 dashboard.render.com , “new”- “web service”

Gpt3.5 api免费白嫖,通过以下几步简单搞定:
1)登录 dashboard.render.com , “new”- “web service”
account_circle
Littlestone ⏩(@sunnywinnersday) 's Twitter Profile Photo

seb They've been doing it every couple weeks, for 2 years now. GPT4 as it came out was incredibly good. GPT4 as it is now is indistinguishable from GPT3.5 when that came out. Use Claude.

@sebjenseb They've been doing it every couple weeks, for 2 years now. GPT4 as it came out was incredibly good. GPT4 as it is now is indistinguishable from GPT3.5 when that came out. Use Claude.
account_circle