| from __future__ import annotations |
|
|
| from typing import Any |
|
|
|
|
| QUALITY_MODE: dict[str, Any] = { |
| "id": "quality", |
| "label": "Quality", |
| "status": "enabled", |
| "text_model_id": "nvidia/NVIDIA-Nemotron-3-Nano-4B-GGUF", |
| "text_runtime": "llama.cpp", |
| "text_format": "GGUF", |
| "text_quantization": "Q4_K_M", |
| "image_model_id": "black-forest-labs/FLUX.2-klein-9B", |
| "image_runtime": "Diffusers", |
| "image_format": "safetensors", |
| "image_precision": "bf16", |
| "image_output_size_px": 1024, |
| "audio_model_id": "openbmb/VoxCPM2", |
| "audio_runtime": "voxcpm", |
| "audio_format": "safetensors", |
| "audio_precision": "bf16", |
| "visible_to_user": True, |
| } |
|
|
| SPEED_MODE: dict[str, Any] = { |
| "id": "speed", |
| "label": "Speed", |
| "status": "disabled_planned", |
| "text_model_id": "nvidia/NVIDIA-Nemotron-3-Nano-4B-GGUF", |
| "text_runtime": "llama.cpp", |
| "text_format": "GGUF", |
| "text_quantization": "Q4_K_M", |
| "image_model_id": "black-forest-labs/FLUX.2-klein-4B", |
| "image_runtime": "Diffusers", |
| "image_format": "safetensors", |
| "image_precision": "bf16", |
| "image_output_size_px": 720, |
| "audio_model_id": "openbmb/VoxCPM2", |
| "audio_runtime": "voxcpm", |
| "audio_format": "safetensors", |
| "audio_precision": "bf16", |
| "visible_to_user": True, |
| } |
|
|
| LOCKED_GENERATION_MODES = [QUALITY_MODE, SPEED_MODE] |
|
|
|
|
| def quality_provenance(*, trace_id: str | None = None, fallback_used: bool | None = True) -> dict[str, Any]: |
| return { |
| "text_model_id": QUALITY_MODE["text_model_id"], |
| "text_runtime": QUALITY_MODE["text_runtime"], |
| "text_format": QUALITY_MODE["text_format"], |
| "text_quantization": QUALITY_MODE["text_quantization"], |
| "image_model_id": QUALITY_MODE["image_model_id"], |
| "image_runtime": QUALITY_MODE["image_runtime"], |
| "image_format": QUALITY_MODE["image_format"], |
| "image_precision": QUALITY_MODE["image_precision"], |
| "audio_model_id": QUALITY_MODE["audio_model_id"], |
| "audio_runtime": QUALITY_MODE["audio_runtime"], |
| "audio_format": QUALITY_MODE["audio_format"], |
| "audio_precision": QUALITY_MODE["audio_precision"], |
| "fallback_used": fallback_used, |
| "trace_id": trace_id, |
| } |
|
|
