ajustes

README.md

@@ -42,7 +42,7 @@ This model is a fine-tuned version of [egonrp/gpt2-wikiwriter-medium-portuguese]
 
 ** It's a chatbot experiment. ;)
 
-The model was trained in 12 hours on a RTX 3060 12GB.
+The model was trained in 12 hours on a RTX 3060 12GB with training argument "--fp16".
 
 
 ## Model description
@@ -59,6 +59,17 @@ More information needed
 
 ## Training procedure
 
+```
+python3 run_clm.py \
+    --model_name_or_path egonrp/gpt2-wikiwriter-medium-portuguese \
+    --train_file /home/egon/dev/gptsquad_data/converted_squad_merged_out_v4c.txt \
+    --do_train \
+    --num_train_epochs 3 \
+    --per_device_train_batch_size 1 \
+    --output_dir /home/egon/dev/gptsquad_model/results_v4c_medium_no_eval \
+    --fp16
+```
+
 ### Training hyperparameters
 
 The following hyperparameters were used during training: