Update model name in L1_3B_ddp.yaml (#1119)

Change 3B config model from `esm2_t33_650M_UR50D` to `esm2_t36_3B_UR50D`

<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->

## Summary by CodeRabbit

* **New Features**
* Updated the training recipe to use the larger ESM2 3B model for
improved capability.
* **Chores**
  * Adjusted experiment tracking run name to align with the new model.
  * Preserved existing batch size and training step settings.

<!-- end of auto-generated comment: release notes by coderabbit.ai -->

Signed-off-by: Jared Wilber <jwilber@nvidia.com>

Author: jwilber

recipes/esm2_native_te_nvfsdp_thd/hydra_config/L1_3B_ddp.yaml

@@ -2,7 +2,7 @@ defaults:
   - defaults
 
 # Training config
-model_name: esm2_t33_650M_UR50D
+model_name: esm2_t36_3B_UR50D
 micro_batch_size: 32
 num_train_steps: 10_000