fixes + improvements

examples/llama-3/diffusion-3.2-1b-pretrain.yaml

@@ -22,8 +22,8 @@ importance_weighting: true
 output_dir: ./outputs/model-out
 
 sequence_len: 512
-sample_packing: true
-eval_sample_packing: true
+sample_packing: false
+eval_sample_packing: false
 
 gradient_accumulation_steps: 8
 micro_batch_size: 4
@@ -51,8 +51,8 @@ eval_steps: 1000
 special_tokens:
   pad_token: "<|end_of_text|>"
 
-wandb_project:
-wandb_entity:
+wandb_project: diffusion-plugin
+wandb_entity: axolotl-ai
 wandb_watch:
 wandb_name:
 wandb_log_model: