Update tiny models for Granite4VisionForConditionalGeneration

chat_template.jinja

@@ -0,0 +1 @@
+{{ bos_token }}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if message['role'] == 'user' %}{{ '[INST] ' + message['content'] + ' [/INST]' }}{% elif message['role'] == 'assistant' %}{{ message['content'] + eos_token}}{% else %}{{ raise_exception('Only user and assistant roles are supported!') }}{% endif %}{% endfor %}

config.json

@@ -0,0 +1,98 @@
+{
+  "architectures": [
+    "Granite4VisionForConditionalGeneration"
+  ],
+  "deepstack_layer_map": [
+    [
+      1,
+      0
+    ]
+  ],
+  "downsample_rate": "1/2",
+  "dtype": "float32",
+  "image_grid_pinpoints": [
+    [
+      8,
+      8
+    ]
+  ],
+  "image_seq_length": 576,
+  "image_token_index": 3,
+  "model_type": "granite4_vision",
+  "pad_token_id": 32001,
+  "projector_dropout": 0.0,
+  "qformer_config": {
+    "attention_probs_dropout_prob": 0.1,
+    "cross_attention_frequency": 1,
+    "encoder_hidden_size": 64,
+    "hidden_act": "gelu",
+    "hidden_dropout_prob": 0.1,
+    "hidden_size": 64,
+    "initializer_range": 0.02,
+    "intermediate_size": 64,
+    "layer_norm_eps": 1e-12,
+    "max_position_embeddings": 2048,
+    "model_type": "blip_2_qformer",
+    "num_attention_heads": 1,
+    "num_hidden_layers": 1,
+    "pad_token_id": 0,
+    "use_qformer_text_input": false,
+    "vocab_size": 30522
+  },
+  "spatial_target_layers": [
+    12,
+    15,
+    18,
+    21
+  ],
+  "spatial_vision_layer": -1,
+  "text_config": {
+    "attention_bias": false,
+    "attention_dropout": 0.0,
+    "attention_multiplier": 1.0,
+    "bos_token_id": 1,
+    "embedding_multiplier": 1.0,
+    "eos_token_id": 2,
+    "hidden_act": "gelu",
+    "hidden_size": 64,
+    "initializer_range": 0.02,
+    "intermediate_size": 64,
+    "logits_scaling": 1.0,
+    "max_position_embeddings": 512,
+    "mlp_bias": false,
+    "model_type": "granite",
+    "num_attention_heads": 2,
+    "num_hidden_layers": 2,
+    "num_key_value_heads": 2,
+    "pad_token_id": 32001,
+    "residual_multiplier": 1.0,
+    "rms_norm_eps": 1e-06,
+    "rope_parameters": {
+      "rope_theta": 10000.0,
+      "rope_type": "default"
+    },
+    "tie_word_embeddings": false,
+    "use_cache": true,
+    "vocab_size": 32002
+  },
+  "tie_word_embeddings": false,
+  "transformers_version": "5.16.0.dev0",
+  "vision_config": {
+    "attention_dropout": 0.0,
+    "hidden_act": "gelu",
+    "hidden_size": 64,
+    "image_size": 8,
+    "initializer_factor": 1.0,
+    "initializer_range": 0.02,
+    "intermediate_size": 64,
+    "layer_norm_eps": 1e-05,
+    "model_type": "clip_vision_model",
+    "num_attention_heads": 2,
+    "num_channels": 3,
+    "num_hidden_layers": 2,
+    "patch_size": 2,
+    "projection_dim": 64
+  },
+  "vision_feature_layer": -1,
+  "vision_feature_select_strategy": "default"
+}

generation_config.json

@@ -0,0 +1,10 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 32001,
+  "transformers_version": "5.16.0.dev0",
+  "use_cache": true
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:73cedf6b9fa1a5da106b4b8a7a7b1a1ce677315bf333c846dc750f2f8570d2ac
+size 17792144

preprocessor_config.json

@@ -0,0 +1,51 @@
+{
+  "crop_size": {
+    "longest_edge": 8,
+    "shortest_edge": 8
+  },
+  "do_center_crop": true,
+  "do_convert_rgb": true,
+  "do_normalize": true,
+  "do_pad": true,
+  "do_rescale": true,
+  "do_resize": true,
+  "image_grid_pinpoints": [
+    [
+      336,
+      672
+    ],
+    [
+      672,
+      336
+    ],
+    [
+      672,
+      672
+    ],
+    [
+      1008,
+      336
+    ],
+    [
+      336,
+      1008
+    ]
+  ],
+  "image_mean": [
+    0.48145466,
+    0.4578275,
+    0.40821073
+  ],
+  "image_processor_type": "LlavaNextImageProcessor",
+  "image_std": [
+    0.26862954,
+    0.26130258,
+    0.27577711
+  ],
+  "resample": 3,
+  "rescale_factor": 0.00392156862745098,
+  "size": {
+    "longest_edge": 8,
+    "shortest_edge": 8
+  }
+}

tokenizer_config.json

@@ -0,0 +1,26 @@
+{
+  "add_prefix_space": null,
+  "backend": "tokenizers",
+  "bos_token": "<s>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "</s>",
+  "image_token": "<image>",
+  "is_local": true,
+  "legacy": true,
+  "local_files_only": false,
+  "max_length": null,
+  "model_max_length": 1000000000000000019884624838656,
+  "model_specific_special_tokens": {
+    "image_token": "<image>"
+  },
+  "pad_to_multiple_of": null,
+  "pad_token": "<pad>",
+  "pad_token_type_id": 0,
+  "padding_side": "left",
+  "processor_class": "LlavaNextProcessor",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}