Explore tweets tagged as #FastChat
#工程师工具 发现一个可以免费跑开源大语言模型的在线地址「FastChat」，支持 Vicuna、Koala、Dolly、ChatGLM、Alpaca、LLaMA，使用下来有一种 GPT3.5 的水平，开源很大程度促进了国��� AI 方案的快速落地，很有一种「你爷爷一开源，我爷爷就创新」那种感觉。 🤖 https://t.co/vQBlsbaHhl
17
161
465
ChatVRMをちょっといじって、ローカル起動したFastChatのOpenAIServerと通信させてみるテスト。モデルは fastchat-t5-3b-v1.0 を利用。さくっと上手くいった。
0
1
4
Mistral-7B is now available at https://t.co/Vzs3OFe5O1 under both the "Chatbot Arena" and "Single Model" tab. Test it yourself! We are glad that our tools (FastChat/Skypilot/vLLM) helped the release of this model! Chatbot Arena now serves over 450 billion parameters for
4
33
239
Vicuna-13B 8bit量子化オプションをつけたら、RTX3090で走った。わーい python3 -m fastchat.serve.cli --model-name ./fastchat/model/vicuna-13b/ --load-8bit #Vicuna13B
0
2
4
1/ We’re teaming up with LMSYS to make Chatbot Arena, Vicuna, Fastchat, and SGLang more accessible to all devs and users. With Hyperbolic’s robust decentralized computing platform, open research organization @lmsysorg can expand the reach of their flagship open-source projects
5
26
224
In his #FastChat video discover how @SSCAdvent is redefining fiduciary technology. https://t.co/FTSR3kdaLK
#WealthManagement
0
0
0
In this #FastChat video, Frank LaRosa, CEO of @ERCBoss, challenges the traditional referral-based growth model by highlighting the critical importance of maintaining a strong online brand. https://t.co/2W2NalfRIB
#WealthManagement
1
0
0
LLM 推理服务部署，经过 cog、vllm、fastchat 等尝试，目前看最完善的应该是 OpenLLM，不仅支持 vllm、ctranslate 后端，还支持lora adapter 部署。同时提供OpenAI 兼容 API以及 Docker 镜像构建。 准备深度测试，并切换生产环境为统一的 OpenLLM 方案。
21
37
229
SWE Arena SWE Arena is an open-source platform that extends FastChat with powerful code execution capabilities, enabling direct evaluation of LLM-generated programs across a wide range of outputs - from simple computations to complex visual interfaces. Building upon the FastChat
4
9
47
FYI I've made a fork of LMSys' fastchat and MT Bench so that it will store the chats that generate the scoring from GPT4 - you can use that version here: https://t.co/pKiV9L48TM
3
8
87
#潮流周刊 第126期 - 买了个球 1. 大语言模型的涌现能力：现象与解释 2. 作为工程师应该如何写作 3. Python - 100 天从新手到大师 4. FastChat：在线跑开源大语言模型 5. 高颜值元素周期表的工具 6. 什么是 Linear 设计风格？ 7. 如何变富：金钱、欲望和控制力 https://t.co/k7qMdpp4WV
4
36
98
Recent LLMs: - StableVicuna 13B - WizardLM 7B - GPT4 based Alpaca 30B - OpenAssistant data llama fine tune 30B - RWKV Raven 14B with EvolInstruct added - Replit Code - FastChat-T5 - GPT4-X-Alpasta-30B -GPT4-X-AlpacaDente-30B - llama-30b-supercot -Chimera-13B - Alpacino-30B
4
43
175
ChatGPTやGoogleのBardに匹敵する精度の日本語対応チャットAI 「Vicuna-13B」のウェイトが公開！ Weight: https://t.co/KdlY2nvYC5 デモ: https://t.co/KoTbH8tjhY
https://t.co/40Tou5EHun
https://t.co/jUggf34Cdg
We are excited to release the weights of Vicuna-13B. 🔥 Run it with a single GPU on your own machine! Get the weights: https://t.co/1OMBfXH0jz Web UI demo: https://t.co/Vzs3OFe5O1 Command line demo: see below
2
59
407
FastChat by @lmsysorg is an awesome platform for training and serving LLMs (Vicuna, Llama 2, and more) 🛠️ The best part: you can use it as a drop-in replacement for OpenAI in @llama_index (or openai-python/other toolkits)! 🔌 Example code: https://t.co/ngjverauKM
5
39
198
In this #FastChat, Chirag Gandhi, Co-founder and CEO of Mili, explores how forward-thinking firms are using Mili to bridge the gap between fragmented data silos and actionable intelligence. https://t.co/oXeCP1jiLq
#WealthManagement #FastChat #Aiforadvisors
2
1
0
Along with Vicuna-v1.5, we also released LongChat-v1.5, based on Llama-2 and 32k context length. You can try it in FastChat or evaluate it in the LongChat repo https://t.co/PXbw1TlwTC!
1
18
77