GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
hemlang
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
hemlang
Model Tree
nbeerbower/Huihui-Qwen3.5-9B-abliterated-TIES
Input Modalities
Output Modalities
Supported Functionality
Measured on hembench with every generated program executed by the real interpreter (Q8_0,
greedy). The comparison point is the sibling ...-TIES-Hemlock-SFT model (the same base
with a lighter Hemlock SFT); "after" is the plain TIES base with this adapter merged:
| level | lighter SFT | this adapter | |
|---|---|---|---|
| L1 | syntax | 6/9 | 7/9 |
| L2 | stdlib | 1/5 | 1/5 |
| L3 | algorithms | 2/7 | 7/7 |
| L4 | systems | 2/7 | 4/7 |
| L5 | translation | 0/5 | 2/5 |
| L6 | debugging | 2/5 | 2/5 |
| weighted overall | 28.5% | 57.1% | |
| programs that fail to run | 21 | 8 |
No measurable general-capability cost: ARC 61.5→62.5, wiki perplexity +1.9%. As with the 27B, stdlib is the weak axis — exact function names and signatures are memorised API knowledge; check calls against the docs.
SFT via Merlina: LoRA r32/α64, dropout 0.05, lr 2e-4 cosine, 2 epochs, bf16, max_length 2048, on a single RTX A6000 (718 steps, final loss 0.090).
The base is a Qwen3.5 VLM whose checkpoint carries 15 mtp.* tensors (the multi-token-
prediction head). peft merge_and_unload() silently drops them, and current llama.cpp then
refuses to load the merged model. Graft them back from the base's model-vlm.safetensors
after merging — they are untouched by a language-model LoRA, so the copy is exact.
Huihui-Qwen3.5-9B-abliterated-TIES also has a sibling with a lighter Hemlock SFT already
baked in (...-TIES-Hemlock-SFT, trained on the 2,673-row base set only — 28.5% hembench).
This adapter was trained on the plain TIES base and supersedes that rung.