---
library_name: transformers
license: mit
base_model: microsoft/Phi-4-reasoning
tags:
  - customer-service
  - supervisor
  - cycleinstruct
  - lg-electronics
  - phi
  - fine-tuned
language:
  - ko
  - en
  - de
  - fr
  - es
pipeline_tag: text-generation
---

# cycleinstruct-phi4-supervisor

Fully merged **microsoft/Phi-4-reasoning** (14.66 B) fine-tuned in two
stages for the LG-Electronics customer-service **quality-supervisor** task.
Given a `(Category, Conversation Transcript, Retrieved Document)` triplet,
the model emits

```
<think>
[Query-Document Alignment] …
[Response-Document Consistency] …
[Response Completeness] …
</think>
{"label": "correct" | "incorrect", "reason": "…"}
```

This repo contains a **single-file, ready-to-use** checkpoint — no adapter
merging required at load time.

## Training pipeline (CycleInstruct-motivated, two-stage SFT)

Following the [CycleInstruct paper](https://arxiv.org/abs/2508.09551)
(EMNLP 2025) as the augmentation strategy motivator:

1. **Stage 1 — CS-chatbot SFT** on 9,868 natural `(question, answer)`
   pairs built from LG feedback + general-inquiry data. LoRA r=16 α=32,
   Muon @ lr=2e-3, seed=1337, 8 epochs.
2. **Stage 2 — Supervisor SFT** on 3,771 human-annotated supervisor
   judgements. Stage-1 LoRA is merged into the base first, then a fresh
   LoRA r=16 α=32 is added and trained with Muon @ lr=1e-3, seed=42,
   7 epochs on 4,096-token sequences.

The uploaded checkpoint is the result of merging **both** LoRA stages into
the base weights and re-saving with `save_pretrained`.

## Metrics — 199-item held-out supervisor test set (T=0, `max_new_tokens=1200`)

| Metric | Stage-1 only | **This model (full merged)** |
|---|---|---|
| Parse-fail rate | 95.98 % | **0.00 %** |
| Accuracy | 1.01 % | **68.84 %** |
| Macro-F1 | 0.033 | **0.615** |
| chrF | 6.55 | **40.92** |
| ROUGE-L | 0.062 | **0.885** |
| BLEU-4 | 0.37 | **22.41** |
| BERTScore-F1 | 0.826 | **0.901** |
| SBERT-cos (multi-mpnet) | 0.437 | **0.830** |

Per-class:

| Class | Precision | Recall | F1 | Support |
|---|---|---|---|---|
| correct   | 0.417 | 0.481 | 0.446 | 52 |
| incorrect | 0.806 | 0.762 | 0.783 | 147 |

## Loading

```python
from transformers import AutoTokenizer, AutoModelForCausalLM
import torch

REPO = "shareit/cycleinstruct-phi4-supervisor"

tok   = AutoTokenizer.from_pretrained(REPO)
model = AutoModelForCausalLM.from_pretrained(
    REPO, torch_dtype=torch.bfloat16,
    attn_implementation="sdpa", device_map="auto").eval()

SYSTEM = "당신은 전자제품 CS 챗봇의 품질을 평가하는 수퍼바이저입니다."
USER   = "[Category] W/M\n[Conversation Transcript] …\n[Retrieved Document] …"

# Phi-4-reasoning ChatML with our clean system prompt (skip default Thought scaffold)
prompt = (
    f"<|im_start|>system<|im_sep|>{SYSTEM}<|im_end|>"
    f"<|im_start|>user<|im_sep|>{USER}<|im_end|>"
    f"<|im_start|>assistant<|im_sep|>"
)
out = model.generate(
    **tok(prompt, return_tensors="pt", add_special_tokens=False).to(model.device),
    do_sample=False, max_new_tokens=1200,
    pad_token_id=tok.pad_token_id,
)
print(tok.decode(out[0], skip_special_tokens=False))
```

`max_new_tokens=1200` matters — the `<think>` block usually consumes
500-900 tokens before the final JSON verdict.

## Training details (stage 2, on top of stage-1-merged base)

- **PEFT**: LoRA r=16, α=32, dropout 0.05, `target_modules=all-linear`, bias='none'
- **Optimizer**: Muon on 2D matrices (Newton-Schulz orthogonalisation) + AdamW on 1D params
- **LR**: 1e-3 (matrix) / 1e-4 (aux), cosine decay with 3 % warmup, grad-clip 1.0
- **Batch**: per-device 1 × grad-accum 16 (effective 16)
- **Seq len**: 4096 (user text char-clipped if exceeds; assistant always preserved)
- **Seed**: 42, **Epochs**: 7
- **Attention**: SDPA (bf16 native on H200)
- **Wall clock**: 5h48m on a half-H200 (48 GB active)

## Data

- Stage-1 train: 9,868 `(q, a)` pairs from `data/processed/train_pairs.jsonl`
  (multilingual, mostly English, ~50 % English, ~15 % German, then FR/ES/IT/JA/ZH…)
- Stage-2 train: 3,771 supervisor-annotated rows
  `{"conversations": [{"from":"system", …}, {"from":"user", …}, {"from":"assistant", …}]}`
  with the assistant response being a `<think>…</think>{"label":…,"reason":…}` judgement.
- Test: 199 held-out supervisor rows (unseen during either stage).

## Intended use / limitations

- Intended for research reproduction of CycleInstruct-style continuation
  training on labeled downstream tasks.
- The `correct` class has substantially lower F1 (0.446) than `incorrect`
  (0.783), reflecting the 39/61 % class imbalance in the training data.
  Class-weighted loss or balanced sampling would likely help.
- The `<think>` reasoning is Korean; input transcripts may be any language.

## License

MIT (inherits from the `microsoft/Phi-4-reasoning` base model).