Add model card
Browse files
README.md
ADDED
|
@@ -0,0 +1,19 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# MultiModel Small 229M
|
| 2 |
+
|
| 3 |
+
Small multimodal-ready transformer checkpoint trained on the project's educational text dataset.
|
| 4 |
+
|
| 5 |
+
## Details
|
| 6 |
+
|
| 7 |
+
- Parameters: approximately 229M
|
| 8 |
+
- Text decoder: 12 layers, 768 hidden size, 12 attention heads
|
| 9 |
+
- Vocabulary: 50,257 GPT-2 BPE tokens
|
| 10 |
+
- Training samples: 1,000
|
| 11 |
+
- Checkpoint: global step 350
|
| 12 |
+
|
| 13 |
+
## Files
|
| 14 |
+
|
| 15 |
+
- `pytorch_model.bin`: PyTorch state dictionary
|
| 16 |
+
- `config.json`: model configuration
|
| 17 |
+
- `tokenizer.json` and `tokenizer_config.json`: uploaded at repository root
|
| 18 |
+
|
| 19 |
+
This checkpoint uses the custom model implementation in the source project and is not directly loadable with `AutoModel`.
|