Standalone RL-MPQ (Reinforcement Learning Mixed-Precision Quantization) checkpoint for the Extreme Survival scenario — a quantized variant of Qwen/Qwen2.5-7B.

Field	Value
Base model	Qwen/Qwen2.5-7B
Scenario	Extreme Survival
Avg bits / weight	2.4643
Compression vs FP16	6.4928×
WikiText-2 PPL	497.4791
Layers	28
Bit distribution	`{'2': 15, '3': 13}`
Format	Fake-quant FP16 + `rlmpq_policy.json`

Usage

python

from transformers import AutoModelForCausalLM, AutoTokenizer

repo = "AvoCahDoe/qwen2-5-7b-rlmpq-extreme-survival"

model = AutoModelForCausalLM.from_pretrained(repo, torch_dtype="float16")
tokenizer = AutoTokenizer.from_pretrained(repo)

Other Qwen 2.5 7B scenarios

Scenario	Avg bits	Compression	WikiText-2 PPL
Aggressive	3.1429	5.0909x	9.3678
Balanced	3.3929	4.7158x	8.9305
Conservative	3.6786	4.3495x	8.4114
High Fidelity	3.75	4.2667x	8.208

Grouped archive (all scenarios in one repo): AvoCahDoe/qwen2-5-7b-rlmpq

Method

Phase 3 — PPO agent assigns per-layer bit widths under the Extreme Survival reward target.
Phase 4 — Policy replayed on real weights; WikiText-2 perplexity validates quality.
Export — Fake-quantized FP16 weights compatible with Hugging Face Transformers.

Files

File	Description
`config.json`	Llama architecture + RL-MPQ metadata
`model.safetensors`	Fake-quantized weights
`rlmpq_policy.json`	Per-layer bit-width policy
`rlmpq_metrics.json`	Validation & PPL summary

Citation

bibtex

@misc{rlmpq_qwen2_5_7b_extreme-survival_2026,
  title  = {RL-MPQ Extreme Survival: Qwen 2.5 7B Mixed-Precision Quantization},
  author = {AvoCahDoe},
  year   = {2026},
  url    = {https://huggingface.co/AvoCahDoe/qwen2-5-7b-rlmpq-extreme-survival}
}