GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
jhone888
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
jhone888
Model Tree
huihui-ai/Huihui-Ornith-1.5-35B-A3B-abliterated
Input Modalities
Output Modalities
Supported Functionality
The full checkpoint was re-ablated with OBLITERATUS 0.1.3 (advanced mode) by elder-plinius:
| Parameter | Value |
|---|---|
| Technique | refusal_direction_ablation |
| Direction method | Whitened SVD (multi-direction) |
| Projection target | Attention projections, all 40 layers |
| Prompt pairs | 32 harmful / 32 harmless (seed 20260826) |
| KL co-optimization | Verified during ablation |
| CoT preservation | Enabled in tooling; no layers required float interpolation |
Config, tokenizer, and chat template are byte-identical to the parent model — only weights changed.
Ornith-1.5-35B-A3B-abliterated-v2-BF16 (this repo) — full-precision base for further tuningOrnith-1.5-35B-A3B-abliterated-v2-NVFP4 — NVIDIA modelopt NVFP4 (MIXED_PRECISION: W4A16 group-16 routed/shared experts, FP8 attention/KV), ~20 GB, 5 shardsThe NVFP4 variant was round-trip validated with FreeToken's own dequantization kernel against this BF16 source: relative error / amax mean 0.0024–0.0044, p99 ≤ 0.035 across routed-expert gate/up/down and shared-expert weights — i.e., quantization noise only, no structural drift.
Qwen3.5-Next-style MoE (qwen3_5_moe): 40 layers, 256 routed experts (top-k A3B ≈ 3B active params), linear attention + full-attention hybrid, 2048 hidden / 512 MoE intermediate.
python
from transformers import AutoModelForCausalLM, AutoTokenizermodel = AutoModelForCausalLM.from_pretrained("<your-name>/Ornith-1.5-35B-A3B-abliterated-v2",torch_dtype="auto", device_map="auto")tok = AutoTokenizer.from_pretrained("<your-name>/Ornith-1.5-35B-A3B-abliterated-v2")