Reinforcement Learning
Diffusers
Safetensors
English
image-quality-assessment
vision-language
image-editing
Instructions to use RobinY99/MR-IQA-2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use RobinY99/MR-IQA-2 with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("RobinY99/MR-IQA-2", dtype=torch.bfloat16, device_map="cuda") prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k" image = pipe(prompt).images[0] - Notebooks
- Google Colab
- Kaggle
| CONDA_SH=/path/to/miniconda3/etc/profile.d/conda.sh | |
| CONDA_ENV_NAME=mr_iqa_actor_judge | |
| ACTOR_PYTHON=/path/to/miniconda3/envs/mr_iqa_actor_judge/bin/python | |
| ACTOR_MODEL_PATH=/path/to/Qwen3.5-4B | |
| NATIVE_ACTOR_MODEL_PATH=/path/to/Qwen3.5-4B | |
| NO_MASK_ACTOR_MODEL_PATH=/path/to/no_mask/checkpoint-30 | |
| MASK_ACTOR_MODEL_PATH=/path/to/mask/checkpoint-30 | |
| ACTOR_PROCESSOR_PATH=/path/to/Qwen3.5-4B | |
| # Prepared initial-Actor experiment identity recorded in mode_matrix.json. | |
| ACTOR_MODEL_TREE_SHA256=eb67981666c5fc926c829b0d2b49b180c46dc6a74c1c8df895834a54c0fe33f3 | |
| TRAIN_IMAGE_ROOT=/path/to/training_images | |
| EVAL_IMAGE_ROOT=/path/to/evaluation_images | |
| DIFFUSERS_VENV=/path/to/editor_venv | |
| DIFFUSERS_MODEL_PATH=/path/to/huggingface_download/editor | |
| DIFFUSERS_WARMUP_IMAGE= | |
| JUDGER_PYTHON=/path/to/miniconda3/envs/mr_iqa_actor_judge/bin/python | |
| JUDGE_MODEL_ID=source-e5-judge-step725 | |
| JUDGE_MODEL_PATH=/path/to/huggingface_download/judge | |
| JUDGE_MANIFEST_PATH=/path/to/huggingface_download/judge/provenance.json | |
| # Full promoted source-checkpoint identity used by the reward/cache protocol. | |
| JUDGE_MODEL_TREE_SHA256=e25415173aacf515e97d5d561c6647a7a84f586061f3a9b2ab3fc079fe21be0a | |
| # Relocatable digest of the 10 public inference files downloaded from the Hub. | |
| JUDGE_MODEL_EXPORT_TREE_SHA256=21b232a1a30dc765f3e7cf16c00fd270e4be354615fea0120e32f975e2777e5c | |
| JUDGER_PROMPT_SCHEMA=e5_training_reasoning_v5 | |
| JUDGE_PROMPT_HASH=fa78a4ccfd2194a2026ff0b6b722bf22b28f8fa060389c57c4adb1618ac280f6 | |
| JUDGER_BACKEND=e5_qwen35_4b_vllm_judge | |
| ORIGINAL_SCORE_CACHE_PATH=/path/to/locally_generated_original_scores.sqlite | |
| ORIGINAL_SCORE_CACHE_SHA256= | |
| ORIGINAL_SCORE_CACHE_EXPECTED_ROW_COUNT=7000 | |
| ORIGINAL_SCORE_CACHE_EXPECTED_SAMPLE_COUNT=7000 | |
| ORIGINAL_SCORE_CACHE_EXPECTED_ACTOR_IDS=source-e5-judge-step725-original-score | |
| ORIGINAL_SCORE_CACHE_PAYLOAD_SCHEMA=vf_original_score_cache_e5_judge_v1 | |
| ORIGINAL_SCORE_CACHE_EXPECTED_RATING_MIN=0.0 | |
| ORIGINAL_SCORE_CACHE_EXPECTED_RATING_MAX=5.0 | |
| FLASH_ATTN_WHEEL=/path/to/validated_flash_attn.whl | |
| FLASH_ATTN_WHEEL_SHA256= | |
| WANDB_MODE=offline | |
| WANDB_PROJECT=mr-iqa-2 | |
| WANDB_ENTITY= | |
| VALIDATION_LOG_WANDB=0 | |
| VF_MIN_FREE_GIB=500 | |