Standalone
RL-MPQ (Reinforcement Learning Mixed-Precision Quantization) checkpoint for the
Aggressive scenario — a quantized variant of
Qwen/Qwen2.5-7B.
1from transformers import AutoModelForCausalLM, AutoTokenizer
2
3repo = "AvoCahDoe/qwen2-5-7b-rlmpq-aggressive"
4
5model = AutoModelForCausalLM.from_pretrained(repo, torch_dtype="float16")
6tokenizer = AutoTokenizer.from_pretrained(repo)
Grouped archive (all scenarios in one repo):
AvoCahDoe/qwen2-5-7b-rlmpq1@misc{rlmpq_qwen2_5_7b_aggressive_2026,
2 title = {RL-MPQ Aggressive: Qwen 2.5 7B Mixed-Precision Quantization},
3 author = {AvoCahDoe},
4 year = {2026},
5 url = {https://huggingface.co/AvoCahDoe/qwen2-5-7b-rlmpq-aggressive}
6}
