Update README.md
Browse files
README.md
CHANGED
|
@@ -1,39 +1,35 @@
|
|
| 1 |
# zjkarina/omniRecsysLLM_idmodality
|
| 2 |
|
| 3 |
-
|
| 4 |
|
| 5 |
-
##
|
| 6 |
|
| 7 |
-
|
| 8 |
|
| 9 |
-
##
|
| 10 |
|
| 11 |
-
-
|
| 12 |
-
-
|
| 13 |
-
-
|
| 14 |
-
-
|
| 15 |
-
-
|
| 16 |
|
| 17 |
## Использование
|
| 18 |
|
| 19 |
```python
|
| 20 |
from any2any_trainer.models.recommendation import RecommendationModel
|
| 21 |
|
| 22 |
-
#
|
| 23 |
model = RecommendationModel.from_pretrained("zjkarina/omniRecsysLLM_idmodality")
|
| 24 |
|
| 25 |
-
#
|
| 26 |
recommendations = model.predict_next_item(
|
| 27 |
-
text="
|
| 28 |
-
id_ids=[12345, 67890], #
|
| 29 |
top_k=5
|
| 30 |
)
|
| 31 |
```
|
| 32 |
|
| 33 |
-
##
|
| 34 |
|
| 35 |
-
|
| 36 |
-
|
| 37 |
-
## Лицензия
|
| 38 |
-
|
| 39 |
-
Apache-2.0 License
|
|
|
|
| 1 |
# zjkarina/omniRecsysLLM_idmodality
|
| 2 |
|
| 3 |
+
Recommendation model based on ID modality for Amazon Fashion.
|
| 4 |
|
| 5 |
+
## Description
|
| 6 |
|
| 7 |
+
This model uses item ID embeddings to generate recommendations based on users’ purchase history.
|
| 8 |
|
| 9 |
+
## Architecture
|
| 10 |
|
| 11 |
+
- **Base model**: Qwen2.5-Omni-7B
|
| 12 |
+
- **Item vocabulary size**: 709,036
|
| 13 |
+
- **ID embedding dimension**: 512
|
| 14 |
+
- **Fusion head dimension**: 1024
|
| 15 |
+
- **Dataset**: Amazon Fashion 2023 Full
|
| 16 |
|
| 17 |
## Использование
|
| 18 |
|
| 19 |
```python
|
| 20 |
from any2any_trainer.models.recommendation import RecommendationModel
|
| 21 |
|
| 22 |
+
# Load model
|
| 23 |
model = RecommendationModel.from_pretrained("zjkarina/omniRecsysLLM_idmodality")
|
| 24 |
|
| 25 |
+
# Generate recommendations
|
| 26 |
recommendations = model.predict_next_item(
|
| 27 |
+
text="The user bought jeans and a t-shirt",
|
| 28 |
+
id_ids=[12345, 67890], # Item IDs from purchase history
|
| 29 |
top_k=5
|
| 30 |
)
|
| 31 |
```
|
| 32 |
|
| 33 |
+
## Training
|
| 34 |
|
| 35 |
+
The model was trained on the Amazon Fashion 2023 dataset using the ID modality.
|
|
|
|
|
|
|
|
|
|
|
|