license: apache-2.0
base_model: 0xSero/Qwen3.5-88B
library_name: gguf
pipeline_tag: text-generation
tags:
- gguf
- abliterated
- qwen3.5
- moe
- reap
- not-for-all-audiences
Qwen3.5-122B-A10B-REAP-30 - Abliterated GGUF
Abliterated, importance-matrix (imatrix) quantized GGUFs of
0xSero/Qwen3.5-88B -
0xSero's ~30% MoE expert-prune (REAP) of Qwen/Qwen3.5-122B-A10B,
taking the model from 122B down to ~88B parameters while keeping the A10B active-expert budget and
the qwen35moe architecture. Robinson Labs then abliterated the pruned model and quantized it here.
Unlike the full 122B sibling, this REAP variant has no Multi-Token Prediction (MTP / NextN): the
upstream config declared a phantom nextn layer carrying no weights, so our convert produced a clean
48-layer model (block_count=48). These are standard single-token-prediction GGUFs.
A full-precision bf16 safetensors base for re-abliteration, LoRA merge, fine-tune, or rolling your
own quants is at RobinsonLabs/Qwen3.5-122B-A10B-REAP-30-abliterated.
Disclosure
This model is abliterated - the hard-refusal reflex on adult / creative content has been
reduced via single-direction weight orthogonalization. Harm guardrails are retained by design:
self-harm prompts still redirect to help (e.g. 988), and it is not intended to assist genuine
wrongdoing. This is a v1, partial abliteration; capability is preserved. Taggednot-for-all-audiences. Use responsibly - you are responsible for your use. License inherited from
the base model: Apache-2.0.
Files
| File | Quant | ~Size | Notes |
|---|---|---|---|
Qwen3.5-122B-A10B-REAP-30-abliterated-Q6_K.gguf |
Q6_K | ~67 GB | near-lossless |
Qwen3.5-122B-A10B-REAP-30-abliterated-Q5_K_M.gguf |
Q5_K_M | ~58 GB | |
Qwen3.5-122B-A10B-REAP-30-abliterated-Q4_K_M.gguf |
Q4_K_M | ~50 GB | |
Qwen3.5-122B-A10B-REAP-30-abliterated-Q4_K_S.gguf |
Q4_K_S | ~47 GB | |
Qwen3.5-122B-A10B-REAP-30-abliterated-IQ4_XS.gguf |
IQ4_XS | ~44 GB | quality/size sweet spot |
Qwen3.5-122B-A10B-REAP-30-abliterated-Q3_K_M.gguf |
Q3_K_M | ~39 GB | |
Qwen3.5-122B-A10B-REAP-30-abliterated-IQ3_M.gguf |
IQ3_M | ~36 GB | |
Qwen3.5-122B-A10B-REAP-30-abliterated-IQ3_XS.gguf |
IQ3_XS | ~34 GB | |
Qwen3.5-122B-A10B-REAP-30-abliterated-IQ2_M.gguf |
IQ2_M | ~27 GB | |
Qwen3.5-122B-A10B-REAP-30-abliterated-IQ2_XS.gguf |
IQ2_XS | ~24 GB | smallest |
All quants are imatrix-weighted (generic calibration). This is the REAP-pruned 48-layer model - no
MTP block.
Method
- Expert prune (REAP): ~30% of the MoE experts removed by 0xSero's REAP method, 122B -> ~88B
params,qwen35moearch, A10B active budget retained. - Abliteration: single mid-layer refusal direction removed via weight orthogonalization on the
bf16 pruned base; routers preserved. No MTP/NextN block exists in this variant. - Quant: importance-matrix (imatrix) weighted convert + quantize with
llama.cpp. The imatrix here uses a generic calibration
set (bartowskicalibration_datav3) - a broad, domain-agnostic fit.
bf16 base
The full-precision bf16 safetensors base for this ladder is
RobinsonLabs/Qwen3.5-122B-A10B-REAP-30-abliterated
- the master for further surgery (re-abliteration, LoRA merge, fine-tune) and for making your own
quants. The upstream REAP parent is 0xSero/Qwen3.5-88B.
Provenance
Qwen3.5-122B-A10B (Apache-2.0) -> REAP-30 expert-prune (0xSero) -> abliterated bf16 (Robinson Labs)
-> Q8_0 master -> generic-imatrix GGUF quants.