Post 190 the training for SnowflakeCore-G1-1B and 7B would be retaken because now I implemented DeepSpeed and management to use two gpus. See translation 👍 1 1 +