limbicnation
ebbb25a2a3
Add PromptDualStreamRefinerNode that produces a positive and negative prompt pair in a single pass via Ollama, intended for the shipped Q8 GGUF of qwen2-5-7b-dual-stream-prompt-lora. Reuses OllamaClient for streaming, timeouts, progress, and llama-runner crash handling, and parses Positive/Negative output defensively across label variants. Includes config/Modelfile.dualstream, unit tests, node registration, and a corrected implementation plan replacing the invalid local transformers/PEFT approach.
40 lines
1.8 KiB
Plaintext
40 lines
1.8 KiB
Plaintext
# Dual-Stream Prompt Refiner Modelfile
|
|
#
|
|
# Registers the pre-merged Q8 GGUF from
|
|
# Limbicnation/qwen2-5-7b-dual-stream-prompt-lora with Ollama so the
|
|
# "Prompt Dual-Stream Refiner" ComfyUI node can use it.
|
|
#
|
|
# Usage:
|
|
# 1. Download qwen2-5-7b-dual-stream-q8.gguf from the HF repo into this dir:
|
|
# huggingface-cli download Limbicnation/qwen2-5-7b-dual-stream-prompt-lora \
|
|
# qwen2-5-7b-dual-stream-q8.gguf --local-dir .
|
|
# 2. Create the Ollama model (name contains "prompt" so it sorts to the top
|
|
# of the node's model dropdown — see OllamaClient.LORA_KEYWORDS):
|
|
# ollama create limbicnation-dualstream-prompt -f config/Modelfile.dualstream
|
|
# 3. Verify the output format before relying on the node's parser:
|
|
# ollama run limbicnation-dualstream-prompt "a mystical forest at twilight"
|
|
# Confirm it emits "Positive: ...\nNegative: ...". If the trained format
|
|
# differs, adjust INSTRUCTION_PROMPT / parse_dual_stream in
|
|
# nodes/prompt_dual_stream_refiner_node.py to match.
|
|
# 4. Restart ComfyUI — the model appears in the node's dropdown.
|
|
|
|
# Pre-merged Q8 GGUF (already contains the LoRA — no ADAPTER needed).
|
|
FROM ./qwen2-5-7b-dual-stream-q8.gguf
|
|
|
|
# Base is Qwen2.5-7B-Instruct, which uses the standard ChatML template.
|
|
# (Ollama applies a sensible Qwen2.5 default; uncomment to pin explicitly.)
|
|
# TEMPLATE """{{ if .System }}<|im_start|>system
|
|
# {{ .System }}<|im_end|>
|
|
# {{ end }}<|im_start|>user
|
|
# {{ .Prompt }}<|im_end|>
|
|
# <|im_start|>assistant
|
|
# """
|
|
|
|
SYSTEM """You are an expert Stable Diffusion prompt engineer. Given a description, you produce a detailed positive prompt and a matching negative prompt, formatted as:
|
|
Positive: <positive prompt>
|
|
Negative: <negative prompt>"""
|
|
|
|
PARAMETER temperature 0.7
|
|
PARAMETER top_p 0.9
|
|
PARAMETER num_ctx 4096
|