Spaces:

openaccess-ai-collective
/

ggml-ui

Build error

ggml-ui / config.yml

fix layout, max size back to 1, llama.cpp doesn't like parallel calls

80c7d2e over 1 year ago

255 Bytes

	---
	repo: TheBloke/wizard-vicuna-13B-GGML
	file: wizard-vicuna-13B.ggml.q5_1.bin
	llama_cpp:
	n_ctx: 2048
	n_gpu_layers: 40 # llama 13b has 40 layers
	chat:
	stop:
	- "</s>"
	- "<unk>"
	- "### User:"
	queue:
	max_size: 16
	concurrency_count: 1