File size: 1,271 Bytes
757b728 f65d1a3 1fa3390 f65d1a3 |
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 |
---
license: gpl-3.0
datasets:
- agentlans/high-quality-english-sentences
language:
- en
base_model:
- unsloth/Qwen3-0.6B-Base
---
# Qwen3 0.6B Text FIM
This is a model trained to do fill-in-the-middle (FIM) with text.\
This model is used in the [browser-autocomplete](https://github.com/OleFranz/browser-autocomplete) extension.\
It was fine tuned using Unsloth, code [here](https://github.com/OleFranz/qwen3-fim-finetune).
## Example code
```python
from transformers import AutoModelForCausalLM, AutoTokenizer
model = AutoModelForCausalLM.from_pretrained("OleFranz/Qwen3-0.6B-Text-FIM")
tokenizer = AutoTokenizer.from_pretrained("OleFranz/Qwen3-0.6B-Text-FIM")
prefix = "do you k"
suffix = " the current time?"
prompt = f"<|fim_prefix|>{prefix}<|fim_suffix|>{suffix}<|fim_middle|>"
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(
**inputs,
max_new_tokens=64,
do_sample=True,
temperature=0.1,
pad_token_id=tokenizer.eos_token_id
)
middle = tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:], skip_special_tokens=True)
GREEN = "\x1b[32m"
RESET = "\x1b[0m"
print(f"Completed text:\n{prefix}{GREEN}{middle}{RESET}{suffix}\n--------------")
raw = tokenizer.decode(outputs[0])
print(f"\nRaw:\n{raw!r}\n---")
``` |