pwipo/tensorflow-chat-gpu repository overview

Image for tensorflow with tensorflow_text and cuda gpu for amd64. Server with ChatGpt Api-like, implement /chat/completions and /models.

tensorflow_text load from https://www.wheelodex.org/projects/tensorflow-text/⁠⁠

---

docker run --env NVIDIA_DISABLE_REQUIRE=1 --gpus all -d -p 8085:8080 --name nn-server pwipo/tensorflow-chat-gpu:1.0.0

work well with Shelf (https://www.shelfmc.ru⁠) desktop apps SimpleChat with SimpleChat::Bot.

batch installation file (.ini):

---

a=SimpleChat::last

a=SimpleChat::Bot::last

s=SimpleChat.Bot::bot.type=custom

s=SimpleChat.Bot::bot.url_prefix=http://localhost:8085⁠

s=SimpleChat.Bot::bot.model=gemma2_instruct_2b_en

s=SimpleChat.Bot::bot.max_tokens=1024

---