Lyell Hintz
patreon
ChatTD 💬 + Dalle 3 + Vision + More [ 1.1.1 ]
🕑 Added 2023-11-14 05:02:40 +0000 UTCUpdated to include a few api endpoints that openai announced last Monday.
🙈 Vision - (must use model 'gpt-4-vision-preview') You can now input an image to chatTD op to be seen by the system along with the message. Top Image parameter will reference and save from a Top reference.
📷 Dalle Image Generation - Integrated DALL-E model for generating images directly from text prompts. Offers options to select model versions (DALL-E 2 or 3), control image resolution, and specify quality settings.
WARNING ⚠⚠⚠ Dalle3 is very pricey and very limited in the amount of parameters compared to something like Stable Diffusion. The quality is great, but be warned $$.
🐉 JSON Mode - Enables JSON mode for API responses. When active, the model is constrained to generate responses that parse into valid JSON objects. Ensure your system message instructs the model to produce JSON, or the API will return an error.
🌱 Seed - Now supports deterministic outputs with Seed parameter. I made -1 be random (essentially doesn't pass any seed value to API)
Other stuff that I am expanding on but is also in this operator:
🛠️ Function Calling (minimally supported / not fully tested yet. more on this coming soon) - Enter a JSON-formatted string defining the functions and their arguments that the model can call. The format should match the OpenAI API specifications for function calls.
(I believe pricing calculations are accurate, but best to refer to your openai account details. You can set a usage limit ! )
![ChatTD 💬 + Dalle 3 + Vision + More [ 1.1.1 ]](https://img10.onlyreon.com/storage/2/qf/qr/32fcc8-019e8036-d3a3-7195-868c-4a427a102fd5.png)
