diff --git a/config/models/large.json b/config/models/large.json index 39bb6ca..f5ee56e 100644 --- a/config/models/large.json +++ b/config/models/large.json @@ -5,5 +5,5 @@ "ff_mult": 4, "drop_p": 0.1, "max_seq_len": 2048, - "grad_checkpoint": false + "grad_checkpoint": true } \ No newline at end of file diff --git a/config/models/xlarge.json b/config/models/xlarge.json index e2c867e..e6c76ab 100644 --- a/config/models/xlarge.json +++ b/config/models/xlarge.json @@ -5,5 +5,5 @@ "ff_mult": 4, "drop_p": 0.1, "max_seq_len": 4096, - "grad_checkpoint": false + "grad_checkpoint": true } \ No newline at end of file