Update README.md
Browse files
README.md
CHANGED
|
@@ -9,17 +9,17 @@ Here you can find the training configs for [SmoLLM3-3B-Base](https://huggingface
|
|
| 9 |
|
| 10 |
The model was trained on 11.2T tokens in 3 stages on 4k context:
|
| 11 |
|
| 12 |
-
- stage 1 config
|
| 13 |
-
- stage 2 config
|
| 14 |
-
- stage 3 config
|
| 15 |
|
| 16 |
|
| 17 |
|
| 18 |
|
| 19 |
And then we trained on an additional 2 stages to extend the contetx length to 64k:
|
| 20 |
|
| 21 |
-
- stage 4 config
|
| 22 |
-
- stage 5 config
|
| 23 |
|
| 24 |
|
| 25 |
|
|
|
|
| 9 |
|
| 10 |
The model was trained on 11.2T tokens in 3 stages on 4k context:
|
| 11 |
|
| 12 |
+
- stage 1 [config](https://huggingface.co/datasets/HuggingFaceTB/smollm3-configs/blob/main/stage1_8T.yaml)
|
| 13 |
+
- stage 2 [config](https://huggingface.co/datasets/HuggingFaceTB/smollm3-configs/blob/main/stage2_8T_9T.yaml)
|
| 14 |
+
- stage 3 [config](https://huggingface.co/datasets/HuggingFaceTB/smollm3-configs/blob/main/stage3_9T_11T.yaml)
|
| 15 |
|
| 16 |
|
| 17 |
|
| 18 |
|
| 19 |
And then we trained on an additional 2 stages to extend the contetx length to 64k:
|
| 20 |
|
| 21 |
+
- stage 4 [config](https://huggingface.co/datasets/HuggingFaceTB/smollm3-configs/blob/main/long_context_4k_to_32k.yaml)
|
| 22 |
+
- stage 5 [config](https://huggingface.co/datasets/HuggingFaceTB/smollm3-configs/blob/main/long_context_32k_to_64.yaml)
|
| 23 |
|
| 24 |
|
| 25 |
|