GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
adamrotmil
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Run this model inference with full control and performance in your environment.
Model Details
Model Provider
adamrotmil
Model Tree
Qwen/Qwen2.5-7B-Instruct
Input Modalities
Output Modalities
Supported Functionality
v3 (current): the Claudish side of every training pair is authored by Claude
Opus per the repo's style guide,
replacing the earlier distillation of the official
ProgramAsWeights translator (whose pairs judged
2.2/5 style, 2.4/5 faithfulness — the v1/v2 quality ceiling). 10.2k pairs survived a
QC filter (rejecting answered-instead-of-restyled and unfaithful rewrites), including
1.2k coherent multi-paragraph documents built from embedding-clustered related texts.
Fine-tuned on Qwen/Qwen2.5-7B-Instruct with LoRA (bf16, r=32, all linear projections, 2 epochs,
max length 2048). The full training set is released as
adamrotmil/claudish-pairs
(CC-BY-NC-4.0, with per-pair QC verdicts). Pipeline code:
claudish-style-adapter.
| Slice | Direction | Meaning | Judge: style | Judge: faithful |
|---|---|---|---|---|
| standard | → Claudish | 0.85 | 2.9 (1.7) | 3.3 (2.9) |
| standard | → English | 0.84 | 4.8 (4.0) | 3.8 (3.2) |
| long (>800 chars) | → Claudish | 0.88 (0.78) | 3.4 (2.0) | 1.8 (1.4) |
| long (>800 chars) | → English | 0.90 (0.80) | 4.0 (3.8) | 2.3 (1.9) |
Every judged metric improved over v2, some sharply — Claudish → English is now genuinely strong. Two limitations remain. Long-input faithfulness is the weak axis (1.8–2.8/5): a 7B model restyling multi-paragraph text still drops or substitutes details, so verify outputs on long documents. And the adapter scores below its own training data (which judged 3.6/5 style, 3.8/5 faithful) — the usual distillation gap. For quality-critical restyling, use the style guide with a capable instruction-following model; it ships in the same repo for exactly this reason.
Plain English (the paragraph above, rendered by this adapter itself — verbatim, unedited):
All judged metrics improved over v2, some sharply. Claudish → English is now genuinely strong. Two limitations remain: 1) Long-input faithfulness (1.8–2.8/5): A 7B model restyling multi-paragraph text still drops or substitutes details, so verify outputs on long documents; and 2) Adapter scores below its own training data (which judged 3.6/5 style, 3.8/5 faithful). For quality-critical restyling, use the style guide with a capable instruction-following model, which ships in the same repo for this reason.
(It dropped one phrase — "the usual distillation gap" — which rather proves the point.)
The good — English → Claudish (style 3, faithful 5):
In: Compare and contrast two potential methods of customer segmentation. Out: Set two candidate customer segmentation methods against each other: where they align, and where they diverge.
The good — Claudish → English (style 5, faithful 5):
In: Sort the items along the alphabetical spine. Out: Organize the items in alphabetical order.
The bad — English → Claudish (style 3, faithful 3): the restyle works but the added imagery ("a seam in the fabric of how we live") edges past what the input states:
In: Six philosophical questions are: - What is the purpose of life? - Are our actions predetermined or do we have free will? ... Out: Six philosophical questions, each one a seam in the fabric of how we live: - What is the purpose of life? - Are our actions predetermined, or do we hold free will? ...
The ugly — a long technical listing (style 2, faithful 1): the prose survives but the model silently corrupted numbers (learning rate 0.001 → 0.01, max epochs 100 → 10). This is the long-input faithfulness limitation in the flesh — never trust the adapter with data-bearing text you don't verify:
In: Configuration Parameters: - Optimizer: Adam - Loss Function: Binary Cross Entropy - Learning rate: 0.001 - Batch size: 32 - Maximum Epochs: 100 ... Out: Configuration Parameters — the knobs that can be tuned: - Optimizer: Adam - Loss Function: Binary Cross Entropy - Learning rate: 0.01 - Batch size: 32 - Maximum Epochs: 10 ...
python
from peft import AutoPeftModelForCausalLMfrom transformers import AutoTokenizermodel = AutoPeftModelForCausalLM.from_pretrained("adamrotmil/claudish-style-adapter", device_map="auto")tokenizer = AutoTokenizer.from_pretrained("adamrotmil/claudish-style-adapter")PROMPT = '''### Instruction:Rewrite the following text in Claudish style while preserving all facts and meaning.### Input:%s### Response:'''inputs = tokenizer(PROMPT % "The tests failed because the DB connection wasn't closed.",return_tensors="pt").to(model.device)output = model.generate(**inputs, max_new_tokens=512, do_sample=False)print(tokenizer.decode(output[0][inputs["input_ids"].shape[1]:], skip_special_tokens=True))
For the reverse direction, use the instruction: "Rewrite the following Claudish text into plain, direct English while preserving all facts and meaning."
Improvements are welcome, in either venue:
01–09 (seed collection → authoring → QC filtering →
training → judged evaluation → publishing), so a better dataset or training recipe
can be carried end-to-end and measured with the same judge.Apache-2.0 all the way down: fine-tunes, quantizations, and derivative adapters are explicitly encouraged. The most valuable open problem is long-input faithfulness (see the table above); a close second is closing the distillation gap to the training data. If you train something better, open an issue — happy to link it here.