Qwen 2.5 Models
Collection
All LoRa trained Qwen2.5 Models • 2 items • Updated
Qwen 2.5-7B Physics-5k
Warning:
- This model is experimental
- Based on the tests i have done, the model tends to halucinate a lot, fact check everything. An improved version of the model is being worked on.
Training
The model was trained on 5k lines of the database camelai/physics.
All database credit to them.
Training parameters
r = 16
alpha = 32
Epochs: 1
Max sequence length: 2048
Official quantizations
| Name | Quant Method | Bits | Size | Description |
|---|---|---|---|---|
| qwen-2.5-physics.Q4_K_M.gguf | Q4_K_M | 4 | 4.68 GB | |
| qwen-2.5-physics.Q6_K.gguf | Q6_K | 6 | 6.25 GB | |
| qwen-2.5-physics.Q8_0.gguf | Q8_0 | 8 | 8.1 GB |
Further quantization may be coming soon.
- Downloads last month
- 60