2500 steps: re-trained with rank-32/alpha-16, typography dataset from CogVLM included

pytorch_lora_weights.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c2173d6e35855e86ccb6a72ac7b330334e829bad8eecb43d2bfa5d813e608943
-size 3314512
+oid sha256:5bc301ffc9e60ddb6d7c5a204e6747721bf8dd3d6dbecca55771a1bfce49f224
+size 13194264