File size: 2,419 Bytes
6bf2c04
 
 
 
ae237d8
 
6bf2c04
 
 
 
 
 
 
 
 
 
 
 
 
 
ae237d8
 
6bf2c04
 
 
ae237d8
6bf2c04
 
 
 
 
ae237d8
 
 
 
6bf2c04
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
---
language: en
license: apache-2.0
base_model: Nanbeige/Nanbeige4.1-3B
datasets:
  - TurkishCodeMan/Nanbeige4.1-3B-Gmail-Tool-Use-Datasets
tags:
  - tool-use
  - gmail
  - function-calling
  - sft
  - dpo
pipeline_tag: text-generation
---

# Nanbeige4.1-3B — Gmail Tool-Use (SFT + DPO)

Fine-tuned version of [Nanbeige/Nanbeige4.1-3B](https://huggingface.co/Nanbeige/Nanbeige4.1-3B)
for Gmail tool-calling tasks using a two-stage training pipeline.

**Training datasets:** [TurkishCodeMan/Nanbeige4.1-3B-Gmail-Tool-Use-Datasets](https://huggingface.co/datasets/TurkishCodeMan/Nanbeige4.1-3B-Gmail-Tool-Use-Datasets)

## Training Pipeline

### Stage 1 — Supervised Fine-Tuning (SFT)
- **Dataset:** 740 multi-turn Gmail agent traces (`sft/traces_chatml_clean.jsonl`)
- **Format:** ChatML with tool_calls (OpenAI function-calling schema)
- **Method:** LoRA r=16, α=32, 7 target modules
- **Result:** loss 0.8464 → 0.1888 · PPL 2.33 → 1.21

### Stage 2 — Direct Preference Optimization (DPO)
- **Dataset:** 3223 preference pairs (`dpo/dpo_dataset.jsonl`) — 3 rejection strategies:
  - `wrong_tool` — incorrect tool selected (~34%)
  - `missing_args` — required arguments omitted (~32%)
  - `bad_answer` — poor final response (~34%)
- **Method:** DPO β=0.1, sigmoid loss, LoRA r=16, `ref_model=None` (PEFT implicit ref)
- **Result:** val_loss=0.000765 · reward accuracy=100% · normalized margin=+0.52

## Supported Tools

| Tool | Description |
|---|---|
| `search_emails` | Search Gmail inbox with filters |
| `read_email` | Read full email content by ID |
| `send_email` | Send a new email |
| `draft_email` | Create a draft |
| `modify_email` | Add/remove labels, mark read/unread |
| `download_attachment` | Download email attachment |

## Usage

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

model = AutoModelForCausalLM.from_pretrained(
    "TurkishCodeMan/Nanbeige4.1-3B-Gmail-Tool-Use",
    torch_dtype=torch.bfloat16,
    trust_remote_code=True,
)
tokenizer = AutoTokenizer.from_pretrained(
    "TurkishCodeMan/Nanbeige4.1-3B-Gmail-Tool-Use",
    trust_remote_code=True,
)
```

## Training Details

| Parameter | Value |
|---|---|
| Base model | Nanbeige/Nanbeige4.1-3B |
| SFT LoRA rank | 16 |
| DPO LoRA rank | 16 |
| DPO β | 0.1 |
| Max length | 2682 tokens |
| GPU | 1× RTX 4090 24GB |
| Framework | TRL 0.22 · Transformers 4.57 · PEFT 0.18 |