ValueError("Prompt exceeds n_ctx") # TODO: Fix about comfyui_vlm_nodes HOT 6 OPEN

Exact same issue here :|

from comfyui_vlm_nodes.

I suspect this is something from llama-cpp-python and not directly from VLM Nodes? My wish is for VLM nodes to error-out sooner because it can hang for a LONG time (10mins+).

I googled and found this exact phrase:
if llama.n_tokens + embed.contents.n_image_pos > llama.n_ctx():
raise ValueError("Prompt exceeds n_ctx") # TODO: Fix
from here:
https://github.com/abetlen/llama-cpp-python/blob/main/llama_cpp/llama_chat_format.py

I hope they can fix soon. Everything was working fine and then I think a few weeks ago, it started throwing this error for models that used to work.

from comfyui_vlm_nodes.