File size: 1,333 Bytes
009cd62 | 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 | # Qwen3 8M Model with Falcon-H1-0.5B-Instruct Tokenizer
## Model Description
This is an 8M parameter Qwen3 model architecture combined with the Falcon-H1-0.5B-Instruct tokenizer.
- **Architecture**: Qwen3 (transformer with Grouped Query Attention, RMS Normalization, Q/K Normalization, RoPE)
- **Tokenizer**: Falcon-H1-0.5B-Instruct
- **Parameters**: 2,183,552
- **Precision**: BF16
- **Format**: SafeTensors
## Configuration
- vocab_size: 32768
- hidden_size: 64
- num_attention_heads: 4
- num_key_value_heads: 2
- num_hidden_layers: 2
- intermediate_size: 160
- head_dim: 16
- max_position_embeddings: 4096
## Usage
```python
from transformers import Qwen3ForCausalLM, AutoTokenizer
model = Qwen3ForCausalLM.from_pretrained("./workspace/qwen3-8m-falcon-tokenizer")
tokenizer = AutoTokenizer.from_pretrained("./workspace/qwen3-8m-falcon-tokenizer")
# Generate text
inputs = tokenizer("Hello, world!", return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=50)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))
```
## Notes
- This model uses the Qwen3 architecture but with Falcon-H1-0.5B-Instruct tokenizer
- The model is initialized with random weights and should be fine-tuned for specific tasks
- Compatible with the Qwen3 model family APIs and interfaces
|