Updated Readme.md
#2
by
Bharatgpt
- opened
README.md
CHANGED
|
@@ -21,6 +21,7 @@ Param1/
|
|
| 21 |
* **Scheduler**: Cosine Annealing
|
| 22 |
* **Learning_rate**: 3e-4 to 3e-6
|
| 23 |
* **Training Setup**: Running on 512 H100 GPUs
|
|
|
|
| 24 |
* **Precision**: bf16-mixed
|
| 25 |
|
| 26 |
* For Pre-Trained Checkpoint (Param 1): https://aikosh.indiaai.gov.in/home/models/details/bharatgen_param_1_indic_scale_bilingual_foundation_model.html
|
|
|
|
| 21 |
* **Scheduler**: Cosine Annealing
|
| 22 |
* **Learning_rate**: 3e-4 to 3e-6
|
| 23 |
* **Training Setup**: Running on 512 H100 GPUs
|
| 24 |
+
* **Framework**: NVIDIA NeMo
|
| 25 |
* **Precision**: bf16-mixed
|
| 26 |
|
| 27 |
* For Pre-Trained Checkpoint (Param 1): https://aikosh.indiaai.gov.in/home/models/details/bharatgen_param_1_indic_scale_bilingual_foundation_model.html
|