Upload nanoVLM using push_to_hub

config.json

@@ -14,21 +14,25 @@
     "lm_rms_eps": 1e-05,
     "lm_re_base": 100000,
     "lm_max_position_embeddings": 8192,
-    "lm_vocab_size": 49152,
+    "lm_base_vocab_size": 49152,
+    "extra_token_amount": 1,
+    "lm_vocab_size": 49153,
     "lm_n_heads": 15,
     "lm_n_kv_heads": 5,
     "lm_dropout": 0.0,
     "lm_n_blocks": 32,
     "lm_attn_scaling": 1.0,
-    "IMAGE_TOKEN_LENGTH": 49,
-    "TOTAL_SEQUENCE_LENGTH": 512,
-    "lm_max_length": 463,
+    "lm_max_length": 512,
     "lm_use_tokens": false,
     "lm_tie_weights": true,
     "lm_model_type": "HuggingFaceTB/SmolLM2-360M-Instruct",
     "lm_tokenizer": "HuggingFaceTB/SmolLM2-360M-Instruct",
     "lm_eos_token_id": 0,
     "mp_pixel_shuffle_factor": 2,
+    "mp_image_token_length": 64,
+    "vlm_extra_tokens": {
+        "image_token": "<|image|>"
+    },
     "vlm_load_backbone_weights": true,
     "vlm_checkpoint_path": "checkpoints",
     "hf_repo_name": "nanoVLM"

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7a62350af2303daf6b292d7cf595b59544574f80d29727805a2bcad18c213dd8
-size 1802502288
+oid sha256:120923186c590bbb003c064163235c28957cf6ca091f753d8ee6f1a89cbad221
+size 1802506128