@huggingface/transformers 3.1.2 → 3.2.1

package/dist/transformers.cjs

@@ -4920,7 +4920,7 @@ class ImageProcessor extends _utils_generic_js__WEBPACK_IMPORTED_MODULE_0__.Call
      * Pad the image by a certain amount.
      * @param {Float32Array} pixelData The pixel data to pad.
      * @param {number[]} imgDims The dimensions of the image (height, width, channels).
-     * @param {{width:number; height:number}|number} padSize The dimensions of the padded image.
+     * @param {{width:number; height:number}|number|'square'} padSize The dimensions of the padded image.
      * @param {Object} options The options for padding.
      * @param {'constant'|'symmetric'} [options.mode='constant'] The type of padding to add.
      * @param {boolean} [options.center=false] Whether to center the image.
@@ -4938,6 +4938,8 @@ class ImageProcessor extends _utils_generic_js__WEBPACK_IMPORTED_MODULE_0__.Call
         if (typeof padSize === 'number') {
             paddedImageWidth = padSize;
             paddedImageHeight = padSize;
+        } else if (padSize === 'square') {
+            paddedImageWidth = paddedImageHeight = Math.max(imageHeight, imageWidth);
         } else {
             paddedImageWidth = padSize.width;
             paddedImageHeight = padSize.height;
@@ -5583,8 +5585,6 @@ function getNormalizedConfig(config) {
         case 'gpt_neox':
         case 'stablelm':
         case 'opt':
-        case 'phi':
-        case 'phi3':
         case 'falcon':
             mapping['num_heads'] = 'num_attention_heads';
             mapping['num_layers'] = 'num_hidden_layers';
@@ -5600,6 +5600,9 @@ function getNormalizedConfig(config) {
         case 'starcoder2':
         case 'qwen2':
         case 'qwen2_vl':
+        case 'phi':
+        case 'phi3':
+        case 'phi3_v':
             mapping['num_heads'] = 'num_key_value_heads';
             mapping['num_layers'] = 'num_hidden_layers';
             mapping['hidden_size'] = 'hidden_size';
@@ -5632,6 +5635,12 @@ function getNormalizedConfig(config) {
             mapping['num_layers'] = 'n_layers';
             mapping['hidden_size'] = 'd_model';
             break;
+        case 'exaone':
+            mapping['num_heads'] = 'num_key_value_heads';
+            mapping['num_layers'] = 'num_layers';
+            mapping['dim_kv'] = 'head_dim';
+            mapping['num_attention_heads'] = 'num_attention_heads';
+            break;
 
         // Encoder-decoder models
         case 't5':
@@ -5673,6 +5682,7 @@ function getNormalizedConfig(config) {
             mapping['encoder_hidden_size'] = mapping['decoder_hidden_size'] = 'd_model';
             break;
         case 'musicgen_decoder':
+        case 'moonshine':
             mapping['num_encoder_layers'] = mapping['num_decoder_layers'] = 'num_hidden_layers';
             mapping['num_encoder_heads'] = mapping['num_decoder_heads'] = 'num_attention_heads';
             mapping['encoder_hidden_size'] = mapping['decoder_hidden_size'] = 'hidden_size';
@@ -5922,7 +5932,7 @@ __webpack_require__.r(__webpack_exports__);
 
 
 
-const VERSION = '3.1.2';
+const VERSION = '3.2.1';
 
 // Check if various APIs are available (depends on environment)
 const IS_BROWSER_ENV = typeof window !== "undefined" && typeof window.document !== "undefined";
@@ -8020,6 +8030,9 @@ __webpack_require__.r(__webpack_exports__);
 /* harmony export */   EsmForTokenClassification: () => (/* binding */ EsmForTokenClassification),
 /* harmony export */   EsmModel: () => (/* binding */ EsmModel),
 /* harmony export */   EsmPreTrainedModel: () => (/* binding */ EsmPreTrainedModel),
+/* harmony export */   ExaoneForCausalLM: () => (/* binding */ ExaoneForCausalLM),
+/* harmony export */   ExaoneModel: () => (/* binding */ ExaoneModel),
+/* harmony export */   ExaonePreTrainedModel: () => (/* binding */ ExaonePreTrainedModel),
 /* harmony export */   FalconForCausalLM: () => (/* binding */ FalconForCausalLM),
 /* harmony export */   FalconModel: () => (/* binding */ FalconModel),
 /* harmony export */   FalconPreTrainedModel: () => (/* binding */ FalconPreTrainedModel),
@@ -8143,7 +8156,15 @@ __webpack_require__.r(__webpack_exports__);
 /* harmony export */   MobileViTV2Model: () => (/* binding */ MobileViTV2Model),
 /* harmony export */   MobileViTV2PreTrainedModel: () => (/* binding */ MobileViTV2PreTrainedModel),
 /* harmony export */   ModelOutput: () => (/* binding */ ModelOutput),
+/* harmony export */   ModernBertForMaskedLM: () => (/* binding */ ModernBertForMaskedLM),
+/* harmony export */   ModernBertForSequenceClassification: () => (/* binding */ ModernBertForSequenceClassification),
+/* harmony export */   ModernBertForTokenClassification: () => (/* binding */ ModernBertForTokenClassification),
+/* harmony export */   ModernBertModel: () => (/* binding */ ModernBertModel),
+/* harmony export */   ModernBertPreTrainedModel: () => (/* binding */ ModernBertPreTrainedModel),
 /* harmony export */   Moondream1ForConditionalGeneration: () => (/* binding */ Moondream1ForConditionalGeneration),
+/* harmony export */   MoonshineForConditionalGeneration: () => (/* binding */ MoonshineForConditionalGeneration),
+/* harmony export */   MoonshineModel: () => (/* binding */ MoonshineModel),
+/* harmony export */   MoonshinePreTrainedModel: () => (/* binding */ MoonshinePreTrainedModel),
 /* harmony export */   MptForCausalLM: () => (/* binding */ MptForCausalLM),
 /* harmony export */   MptModel: () => (/* binding */ MptModel),
 /* harmony export */   MptPreTrainedModel: () => (/* binding */ MptPreTrainedModel),
@@ -8184,6 +8205,8 @@ __webpack_require__.r(__webpack_exports__);
 /* harmony export */   Phi3ForCausalLM: () => (/* binding */ Phi3ForCausalLM),
 /* harmony export */   Phi3Model: () => (/* binding */ Phi3Model),
 /* harmony export */   Phi3PreTrainedModel: () => (/* binding */ Phi3PreTrainedModel),
+/* harmony export */   Phi3VForCausalLM: () => (/* binding */ Phi3VForCausalLM),
+/* harmony export */   Phi3VPreTrainedModel: () => (/* binding */ Phi3VPreTrainedModel),
 /* harmony export */   PhiForCausalLM: () => (/* binding */ PhiForCausalLM),
 /* harmony export */   PhiModel: () => (/* binding */ PhiModel),
 /* harmony export */   PhiPreTrainedModel: () => (/* binding */ PhiPreTrainedModel),
@@ -8429,6 +8452,7 @@ const MODEL_TYPES = {
     ImageTextToText: 6,
     Musicgen: 7,
     MultiModality: 8,
+    Phi3V: 9,
 }
 //////////////////////////////////////////////////
 
@@ -9204,6 +9228,10 @@ class PreTrainedModel extends _utils_generic_js__WEBPACK_IMPORTED_MODULE_3__.Cal
                 this._forward = imageTextToTextForward;
                 this._prepare_inputs_for_generation = image_text_to_text_prepare_inputs_for_generation;
                 break;
+            case MODEL_TYPES.Phi3V:
+                this.can_generate = true;
+                this._prepare_inputs_for_generation = image_text_to_text_prepare_inputs_for_generation;
+                break;
 
             case MODEL_TYPES.MultiModality:
                 this.can_generate = true;
@@ -9368,6 +9396,18 @@ class PreTrainedModel extends _utils_generic_js__WEBPACK_IMPORTED_MODULE_3__.Cal
                 }, options),
             ]);
 
+        } else if (modelType === MODEL_TYPES.Phi3V) {
+            info = await Promise.all([
+                constructSessions(pretrained_model_name_or_path, {
+                    prepare_inputs_embeds: 'prepare_inputs_embeds',
+                    model: 'model',
+                    vision_encoder: 'vision_encoder',
+                }, options),
+                getOptionalConfigs(pretrained_model_name_or_path, {
+                    generation_config: 'generation_config.json',
+                }, options),
+            ]);
+
         } else { // should be MODEL_TYPES.EncoderOnly
             if (modelType !== MODEL_TYPES.EncoderOnly) {
                 const type = modelName ?? config?.model_type;
@@ -10232,6 +10272,49 @@ class BertForQuestionAnswering extends BertPreTrainedModel {
 }
 //////////////////////////////////////////////////
 
+//////////////////////////////////////////////////
+// ModernBert models
+class ModernBertPreTrainedModel extends PreTrainedModel { }
+class ModernBertModel extends ModernBertPreTrainedModel { }
+
+class ModernBertForMaskedLM extends ModernBertPreTrainedModel {
+    /**
+     * Calls the model on new inputs.
+     *
+     * @param {Object} model_inputs The inputs to the model.
+     * @returns {Promise<MaskedLMOutput>} An object containing the model's output logits for masked language modeling.
+     */
+    async _call(model_inputs) {
+        return new MaskedLMOutput(await super._call(model_inputs));
+    }
+}
+
+class ModernBertForSequenceClassification extends ModernBertPreTrainedModel {
+    /**
+     * Calls the model on new inputs.
+     *
+     * @param {Object} model_inputs The inputs to the model.
+     * @returns {Promise<SequenceClassifierOutput>} An object containing the model's output logits for sequence classification.
+     */
+    async _call(model_inputs) {
+        return new SequenceClassifierOutput(await super._call(model_inputs));
+    }
+}
+
+class ModernBertForTokenClassification extends ModernBertPreTrainedModel {
+    /**
+     * Calls the model on new inputs.
+     *
+     * @param {Object} model_inputs The inputs to the model.
+     * @returns {Promise<TokenClassifierOutput>} An object containing the model's output logits for token classification.
+     */
+    async _call(model_inputs) {
+        return new TokenClassifierOutput(await super._call(model_inputs));
+    }
+}
+//////////////////////////////////////////////////
+
+
 //////////////////////////////////////////////////
 // NomicBert models
 class NomicBertPreTrainedModel extends PreTrainedModel { }
@@ -11640,6 +11723,29 @@ class WhisperForConditionalGeneration extends WhisperPreTrainedModel {
 }
 //////////////////////////////////////////////////
 
+
+//////////////////////////////////////////////////
+// Moonshine models
+class MoonshinePreTrainedModel extends PreTrainedModel {
+
+    requires_attention_mask = false;
+    main_input_name = 'input_values';
+    forward_params = [
+        'input_values',
+        'decoder_input_ids',
+        'past_key_values',
+    ];
+};
+
+/**
+ * MoonshineModel class for training Moonshine models without a language model head.
+ */
+class MoonshineModel extends MoonshinePreTrainedModel { }
+
+class MoonshineForConditionalGeneration extends MoonshinePreTrainedModel { } 
+//////////////////////////////////////////////////
+
+
 //////////////////////////////////////////////////
 /**
  * Vision Encoder-Decoder model based on OpenAI's GPT architecture for image captioning and other vision tasks
@@ -11910,6 +12016,77 @@ class Idefics3ForConditionalGeneration extends Idefics3PreTrainedModel {
 }
 //////////////////////////////////////////////////
 
+class Phi3VPreTrainedModel extends PreTrainedModel {
+    forward_params = [
+        'input_ids',
+        'inputs_embeds',
+        'attention_mask',
+        'position_ids',
+        'pixel_values',
+        'image_sizes',
+        'past_key_values',
+    ];
+}
+class Phi3VForCausalLM extends Phi3VPreTrainedModel {
+
+    async forward({
+        // Produced by the tokenizer/processor:
+        input_ids = null,
+        attention_mask = null,
+        pixel_values = null,
+        image_sizes = null,
+
+        // Used during generation:
+        position_ids = null,
+        inputs_embeds = null,
+        past_key_values = null,
+
+        // Generic generation parameters
+        generation_config = null,
+        logits_processor = null,
+
+        // TODO: needed?
+        ...kwargs
+    }) {
+        if (!inputs_embeds) {
+            let image_features;
+            if (pixel_values && input_ids.dims[1] !== 1) {
+                if (!image_sizes) {
+                    throw new Error('`image_sizes` must be provided when `pixel_values` is provided.');
+                }
+
+                // Encode the image
+                ({ image_features } = await sessionRun(this.sessions['vision_encoder'], {
+                    pixel_values,
+                    image_sizes,
+                }));
+            } else {
+                const hidden_size = this.config.normalized_config.hidden_size;
+                image_features = new _utils_tensor_js__WEBPACK_IMPORTED_MODULE_9__.Tensor(
+                    'float32',
+                    [],
+                    [0, hidden_size],
+                );
+            }
+
+            ({ inputs_embeds } = await sessionRun(this.sessions['prepare_inputs_embeds'], {
+                input_ids,
+                image_features,
+            }));
+        }
+
+        const outputs = await decoderForward(this, {
+            inputs_embeds,
+            past_key_values,
+            attention_mask,
+            position_ids,
+            generation_config,
+            logits_processor,
+        }, false);
+        return outputs;
+    }
+}
+
 //////////////////////////////////////////////////
 class CLIPPreTrainedModel extends PreTrainedModel { }
 
@@ -11964,9 +12141,11 @@ class CLIPModel extends CLIPPreTrainedModel { }
 class CLIPTextModel extends CLIPPreTrainedModel {
     /** @type {typeof PreTrainedModel.from_pretrained} */
     static async from_pretrained(pretrained_model_name_or_path, options = {}) {
-        // Update default model file name if not provided
-        options.model_file_name ??= 'text_model';
-        return super.from_pretrained(pretrained_model_name_or_path, options);
+        return super.from_pretrained(pretrained_model_name_or_path, {
+            // Update default model file name if not provided
+            model_file_name: 'text_model',
+            ...options,
+        });
     }
 }
 
@@ -11999,9 +12178,11 @@ class CLIPTextModel extends CLIPPreTrainedModel {
 class CLIPTextModelWithProjection extends CLIPPreTrainedModel {
     /** @type {typeof PreTrainedModel.from_pretrained} */
     static async from_pretrained(pretrained_model_name_or_path, options = {}) {
-        // Update default model file name if not provided
-        options.model_file_name ??= 'text_model';
-        return super.from_pretrained(pretrained_model_name_or_path, options);
+        return super.from_pretrained(pretrained_model_name_or_path, {
+            // Update default model file name if not provided
+            model_file_name: 'text_model',
+            ...options,
+        });
     }
 }
 
@@ -12011,9 +12192,11 @@ class CLIPTextModelWithProjection extends CLIPPreTrainedModel {
 class CLIPVisionModel extends CLIPPreTrainedModel {
     /** @type {typeof PreTrainedModel.from_pretrained} */
     static async from_pretrained(pretrained_model_name_or_path, options = {}) {
-        // Update default model file name if not provided
-        options.model_file_name ??= 'vision_model';
-        return super.from_pretrained(pretrained_model_name_or_path, options);
+        return super.from_pretrained(pretrained_model_name_or_path, {
+            // Update default model file name if not provided
+            model_file_name: 'vision_model',
+            ...options,
+        });
     }
 }
 
@@ -12046,9 +12229,11 @@ class CLIPVisionModel extends CLIPPreTrainedModel {
 class CLIPVisionModelWithProjection extends CLIPPreTrainedModel {
     /** @type {typeof PreTrainedModel.from_pretrained} */
     static async from_pretrained(pretrained_model_name_or_path, options = {}) {
-        // Update default model file name if not provided
-        options.model_file_name ??= 'vision_model';
-        return super.from_pretrained(pretrained_model_name_or_path, options);
+        return super.from_pretrained(pretrained_model_name_or_path, {
+            // Update default model file name if not provided
+            model_file_name: 'vision_model',
+            ...options,
+        });
     }
 }
 //////////////////////////////////////////////////
@@ -12132,9 +12317,11 @@ class SiglipModel extends SiglipPreTrainedModel { }
 class SiglipTextModel extends SiglipPreTrainedModel {
     /** @type {typeof PreTrainedModel.from_pretrained} */
     static async from_pretrained(pretrained_model_name_or_path, options = {}) {
-        // Update default model file name if not provided
-        options.model_file_name ??= 'text_model';
-        return super.from_pretrained(pretrained_model_name_or_path, options);
+        return super.from_pretrained(pretrained_model_name_or_path, {
+            // Update default model file name if not provided
+            model_file_name: 'text_model',
+            ...options,
+        });
     }
 }
 
@@ -12167,9 +12354,11 @@ class SiglipTextModel extends SiglipPreTrainedModel {
 class SiglipVisionModel extends CLIPPreTrainedModel {
     /** @type {typeof PreTrainedModel.from_pretrained} */
     static async from_pretrained(pretrained_model_name_or_path, options = {}) {
-        // Update default model file name if not provided
-        options.model_file_name ??= 'vision_model';
-        return super.from_pretrained(pretrained_model_name_or_path, options);
+        return super.from_pretrained(pretrained_model_name_or_path, {
+            // Update default model file name if not provided
+            model_file_name: 'vision_model',
+            ...options,
+        });
     }
 }
 //////////////////////////////////////////////////
@@ -12224,18 +12413,22 @@ class JinaCLIPModel extends JinaCLIPPreTrainedModel {
 class JinaCLIPTextModel extends JinaCLIPPreTrainedModel {
     /** @type {typeof PreTrainedModel.from_pretrained} */
     static async from_pretrained(pretrained_model_name_or_path, options = {}) {
-        // Update default model file name if not provided
-        options.model_file_name ??= 'text_model';
-        return super.from_pretrained(pretrained_model_name_or_path, options);
+        return super.from_pretrained(pretrained_model_name_or_path, {
+            // Update default model file name if not provided
+            model_file_name: 'text_model',
+            ...options,
+        });
     }
 }
 
 class JinaCLIPVisionModel extends JinaCLIPPreTrainedModel {
     /** @type {typeof PreTrainedModel.from_pretrained} */
     static async from_pretrained(pretrained_model_name_or_path, options = {}) {
-        // Update default model file name if not provided
-        options.model_file_name ??= 'vision_model';
-        return super.from_pretrained(pretrained_model_name_or_path, options);
+        return super.from_pretrained(pretrained_model_name_or_path, {
+            // Update default model file name if not provided
+            model_file_name: 'vision_model',
+            ...options,
+        });
     }
 }
 //////////////////////////////////////////////////
@@ -12395,6 +12588,14 @@ class LlamaForCausalLM extends LlamaPreTrainedModel { }
 //////////////////////////////////////////////////
 
 
+//////////////////////////////////////////////////
+// EXAONE models
+class ExaonePreTrainedModel extends PreTrainedModel { }
+class ExaoneModel extends ExaonePreTrainedModel { }
+class ExaoneForCausalLM extends ExaonePreTrainedModel { }
+//////////////////////////////////////////////////
+
+
 //////////////////////////////////////////////////
 // MobileLLM models
 class MobileLLMPreTrainedModel extends PreTrainedModel { }
@@ -14457,9 +14658,11 @@ class ClapModel extends ClapPreTrainedModel { }
 class ClapTextModelWithProjection extends ClapPreTrainedModel {
     /** @type {typeof PreTrainedModel.from_pretrained} */
     static async from_pretrained(pretrained_model_name_or_path, options = {}) {
-        // Update default model file name if not provided
-        options.model_file_name ??= 'text_model';
-        return super.from_pretrained(pretrained_model_name_or_path, options);
+        return super.from_pretrained(pretrained_model_name_or_path, {
+            // Update default model file name if not provided
+            model_file_name: 'text_model',
+            ...options,
+        });
     }
 }
 
@@ -14492,9 +14695,11 @@ class ClapTextModelWithProjection extends ClapPreTrainedModel {
 class ClapAudioModelWithProjection extends ClapPreTrainedModel {
     /** @type {typeof PreTrainedModel.from_pretrained} */
     static async from_pretrained(pretrained_model_name_or_path, options = {}) {
-        // Update default model file name if not provided
-        options.model_file_name ??= 'audio_model';
-        return super.from_pretrained(pretrained_model_name_or_path, options);
+        return super.from_pretrained(pretrained_model_name_or_path, {
+            // Update default model file name if not provided
+            model_file_name: 'audio_model',
+            ...options,
+        });
     }
 }
 //////////////////////////////////////////////////
@@ -15080,6 +15285,7 @@ class PretrainedMixin {
 
 const MODEL_MAPPING_NAMES_ENCODER_ONLY = new Map([
     ['bert', ['BertModel', BertModel]],
+    ['modernbert', ['ModernBertModel', ModernBertModel]],
     ['nomic_bert', ['NomicBertModel', NomicBertModel]],
     ['roformer', ['RoFormerModel', RoFormerModel]],
     ['electra', ['ElectraModel', ElectraModel]],
@@ -15181,6 +15387,7 @@ const MODEL_MAPPING_NAMES_DECODER_ONLY = new Map([
     ['gpt_neox', ['GPTNeoXModel', GPTNeoXModel]],
     ['codegen', ['CodeGenModel', CodeGenModel]],
     ['llama', ['LlamaModel', LlamaModel]],
+    ['exaone', ['ExaoneModel', ExaoneModel]],
     ['olmo', ['OlmoModel', OlmoModel]],
     ['olmo2', ['Olmo2Model', Olmo2Model]],
     ['mobilellm', ['MobileLLMModel', MobileLLMModel]],
@@ -15203,6 +15410,7 @@ const MODEL_MAPPING_NAMES_DECODER_ONLY = new Map([
 const MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING_NAMES = new Map([
     ['speecht5', ['SpeechT5ForSpeechToText', SpeechT5ForSpeechToText]],
     ['whisper', ['WhisperForConditionalGeneration', WhisperForConditionalGeneration]],
+    ['moonshine', ['MoonshineForConditionalGeneration', MoonshineForConditionalGeneration]],
 ]);
 
 const MODEL_FOR_TEXT_TO_SPECTROGRAM_MAPPING_NAMES = new Map([
@@ -15216,6 +15424,7 @@ const MODEL_FOR_TEXT_TO_WAVEFORM_MAPPING_NAMES = new Map([
 
 const MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING_NAMES = new Map([
     ['bert', ['BertForSequenceClassification', BertForSequenceClassification]],
+    ['modernbert', ['ModernBertForSequenceClassification', ModernBertForSequenceClassification]],
     ['roformer', ['RoFormerForSequenceClassification', RoFormerForSequenceClassification]],
     ['electra', ['ElectraForSequenceClassification', ElectraForSequenceClassification]],
     ['esm', ['EsmForSequenceClassification', EsmForSequenceClassification]],
@@ -15237,6 +15446,7 @@ const MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING_NAMES = new Map([
 
 const MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING_NAMES = new Map([
     ['bert', ['BertForTokenClassification', BertForTokenClassification]],
+    ['modernbert', ['ModernBertForTokenClassification', ModernBertForTokenClassification]],
     ['roformer', ['RoFormerForTokenClassification', RoFormerForTokenClassification]],
     ['electra', ['ElectraForTokenClassification', ElectraForTokenClassification]],
     ['esm', ['EsmForTokenClassification', EsmForTokenClassification]],
@@ -15273,6 +15483,7 @@ const MODEL_FOR_CAUSAL_LM_MAPPING_NAMES = new Map([
     ['gpt_neox', ['GPTNeoXForCausalLM', GPTNeoXForCausalLM]],
     ['codegen', ['CodeGenForCausalLM', CodeGenForCausalLM]],
     ['llama', ['LlamaForCausalLM', LlamaForCausalLM]],
+    ['exaone', ['ExaoneForCausalLM', ExaoneForCausalLM]],
     ['olmo', ['OlmoForCausalLM', OlmoForCausalLM]],
     ['olmo2', ['Olmo2ForCausalLM', Olmo2ForCausalLM]],
     ['mobilellm', ['MobileLLMForCausalLM', MobileLLMForCausalLM]],
@@ -15292,6 +15503,9 @@ const MODEL_FOR_CAUSAL_LM_MAPPING_NAMES = new Map([
     ['falcon', ['FalconForCausalLM', FalconForCausalLM]],
     ['trocr', ['TrOCRForCausalLM', TrOCRForCausalLM]],
     ['stablelm', ['StableLmForCausalLM', StableLmForCausalLM]],
+
+    // Also image-text-to-text
+    ['phi3_v', ['Phi3VForCausalLM', Phi3VForCausalLM]],
 ]);
 
 const MODEL_FOR_MULTIMODALITY_MAPPING_NAMES = new Map([
@@ -15301,6 +15515,7 @@ const MODEL_FOR_MULTIMODALITY_MAPPING_NAMES = new Map([
 
 const MODEL_FOR_MASKED_LM_MAPPING_NAMES = new Map([
     ['bert', ['BertForMaskedLM', BertForMaskedLM]],
+    ['modernbert', ['ModernBertForMaskedLM', ModernBertForMaskedLM]],
     ['roformer', ['RoFormerForMaskedLM', RoFormerForMaskedLM]],
     ['electra', ['ElectraForMaskedLM', ElectraForMaskedLM]],
     ['esm', ['EsmForMaskedLM', EsmForMaskedLM]],
@@ -15529,6 +15744,7 @@ const CUSTOM_MAPPING = [
     // OVERRIDE:
     // TODO: Refactor to allow class to specify model
     ['MusicgenForConditionalGeneration', MusicgenForConditionalGeneration, MODEL_TYPES.Musicgen],
+    ['Phi3VForCausalLM', Phi3VForCausalLM, MODEL_TYPES.Phi3V],
 
     ['CLIPTextModelWithProjection', CLIPTextModelWithProjection, MODEL_TYPES.EncoderOnly],
     ['SiglipTextModel', SiglipTextModel, MODEL_TYPES.EncoderOnly],
@@ -16783,23 +16999,26 @@ __webpack_require__.r(__webpack_exports__);
 /* harmony export */ __webpack_require__.d(__webpack_exports__, {
 /* harmony export */   ASTFeatureExtractor: () => (/* reexport safe */ _audio_spectrogram_transformer_feature_extraction_audio_spectrogram_transformer_js__WEBPACK_IMPORTED_MODULE_0__.ASTFeatureExtractor),
 /* harmony export */   ClapFeatureExtractor: () => (/* reexport safe */ _clap_feature_extraction_clap_js__WEBPACK_IMPORTED_MODULE_1__.ClapFeatureExtractor),
-/* harmony export */   ImageFeatureExtractor: () => (/* reexport safe */ _base_image_processors_utils_js__WEBPACK_IMPORTED_MODULE_8__.ImageProcessor),
-/* harmony export */   PyAnnoteFeatureExtractor: () => (/* reexport safe */ _pyannote_feature_extraction_pyannote_js__WEBPACK_IMPORTED_MODULE_2__.PyAnnoteFeatureExtractor),
-/* harmony export */   SeamlessM4TFeatureExtractor: () => (/* reexport safe */ _seamless_m4t_feature_extraction_seamless_m4t_js__WEBPACK_IMPORTED_MODULE_3__.SeamlessM4TFeatureExtractor),
-/* harmony export */   SpeechT5FeatureExtractor: () => (/* reexport safe */ _speecht5_feature_extraction_speecht5_js__WEBPACK_IMPORTED_MODULE_4__.SpeechT5FeatureExtractor),
-/* harmony export */   Wav2Vec2FeatureExtractor: () => (/* reexport safe */ _wav2vec2_feature_extraction_wav2vec2_js__WEBPACK_IMPORTED_MODULE_5__.Wav2Vec2FeatureExtractor),
-/* harmony export */   WeSpeakerFeatureExtractor: () => (/* reexport safe */ _wespeaker_feature_extraction_wespeaker_js__WEBPACK_IMPORTED_MODULE_6__.WeSpeakerFeatureExtractor),
-/* harmony export */   WhisperFeatureExtractor: () => (/* reexport safe */ _whisper_feature_extraction_whisper_js__WEBPACK_IMPORTED_MODULE_7__.WhisperFeatureExtractor)
+/* harmony export */   ImageFeatureExtractor: () => (/* reexport safe */ _base_image_processors_utils_js__WEBPACK_IMPORTED_MODULE_9__.ImageProcessor),
+/* harmony export */   MoonshineFeatureExtractor: () => (/* reexport safe */ _moonshine_feature_extraction_moonshine_js__WEBPACK_IMPORTED_MODULE_2__.MoonshineFeatureExtractor),
+/* harmony export */   PyAnnoteFeatureExtractor: () => (/* reexport safe */ _pyannote_feature_extraction_pyannote_js__WEBPACK_IMPORTED_MODULE_3__.PyAnnoteFeatureExtractor),
+/* harmony export */   SeamlessM4TFeatureExtractor: () => (/* reexport safe */ _seamless_m4t_feature_extraction_seamless_m4t_js__WEBPACK_IMPORTED_MODULE_4__.SeamlessM4TFeatureExtractor),
+/* harmony export */   SpeechT5FeatureExtractor: () => (/* reexport safe */ _speecht5_feature_extraction_speecht5_js__WEBPACK_IMPORTED_MODULE_5__.SpeechT5FeatureExtractor),
+/* harmony export */   Wav2Vec2FeatureExtractor: () => (/* reexport safe */ _wav2vec2_feature_extraction_wav2vec2_js__WEBPACK_IMPORTED_MODULE_6__.Wav2Vec2FeatureExtractor),
+/* harmony export */   WeSpeakerFeatureExtractor: () => (/* reexport safe */ _wespeaker_feature_extraction_wespeaker_js__WEBPACK_IMPORTED_MODULE_7__.WeSpeakerFeatureExtractor),
+/* harmony export */   WhisperFeatureExtractor: () => (/* reexport safe */ _whisper_feature_extraction_whisper_js__WEBPACK_IMPORTED_MODULE_8__.WhisperFeatureExtractor)
 /* harmony export */ });
 /* harmony import */ var _audio_spectrogram_transformer_feature_extraction_audio_spectrogram_transformer_js__WEBPACK_IMPORTED_MODULE_0__ = __webpack_require__(/*! ./audio_spectrogram_transformer/feature_extraction_audio_spectrogram_transformer.js */ "./src/models/audio_spectrogram_transformer/feature_extraction_audio_spectrogram_transformer.js");
 /* harmony import */ var _clap_feature_extraction_clap_js__WEBPACK_IMPORTED_MODULE_1__ = __webpack_require__(/*! ./clap/feature_extraction_clap.js */ "./src/models/clap/feature_extraction_clap.js");
-/* harmony import */ var _pyannote_feature_extraction_pyannote_js__WEBPACK_IMPORTED_MODULE_2__ = __webpack_require__(/*! ./pyannote/feature_extraction_pyannote.js */ "./src/models/pyannote/feature_extraction_pyannote.js");
-/* harmony import */ var _seamless_m4t_feature_extraction_seamless_m4t_js__WEBPACK_IMPORTED_MODULE_3__ = __webpack_require__(/*! ./seamless_m4t/feature_extraction_seamless_m4t.js */ "./src/models/seamless_m4t/feature_extraction_seamless_m4t.js");
-/* harmony import */ var _speecht5_feature_extraction_speecht5_js__WEBPACK_IMPORTED_MODULE_4__ = __webpack_require__(/*! ./speecht5/feature_extraction_speecht5.js */ "./src/models/speecht5/feature_extraction_speecht5.js");
-/* harmony import */ var _wav2vec2_feature_extraction_wav2vec2_js__WEBPACK_IMPORTED_MODULE_5__ = __webpack_require__(/*! ./wav2vec2/feature_extraction_wav2vec2.js */ "./src/models/wav2vec2/feature_extraction_wav2vec2.js");
-/* harmony import */ var _wespeaker_feature_extraction_wespeaker_js__WEBPACK_IMPORTED_MODULE_6__ = __webpack_require__(/*! ./wespeaker/feature_extraction_wespeaker.js */ "./src/models/wespeaker/feature_extraction_wespeaker.js");
-/* harmony import */ var _whisper_feature_extraction_whisper_js__WEBPACK_IMPORTED_MODULE_7__ = __webpack_require__(/*! ./whisper/feature_extraction_whisper.js */ "./src/models/whisper/feature_extraction_whisper.js");
-/* harmony import */ var _base_image_processors_utils_js__WEBPACK_IMPORTED_MODULE_8__ = __webpack_require__(/*! ../base/image_processors_utils.js */ "./src/base/image_processors_utils.js");
+/* harmony import */ var _moonshine_feature_extraction_moonshine_js__WEBPACK_IMPORTED_MODULE_2__ = __webpack_require__(/*! ./moonshine/feature_extraction_moonshine.js */ "./src/models/moonshine/feature_extraction_moonshine.js");
+/* harmony import */ var _pyannote_feature_extraction_pyannote_js__WEBPACK_IMPORTED_MODULE_3__ = __webpack_require__(/*! ./pyannote/feature_extraction_pyannote.js */ "./src/models/pyannote/feature_extraction_pyannote.js");
+/* harmony import */ var _seamless_m4t_feature_extraction_seamless_m4t_js__WEBPACK_IMPORTED_MODULE_4__ = __webpack_require__(/*! ./seamless_m4t/feature_extraction_seamless_m4t.js */ "./src/models/seamless_m4t/feature_extraction_seamless_m4t.js");
+/* harmony import */ var _speecht5_feature_extraction_speecht5_js__WEBPACK_IMPORTED_MODULE_5__ = __webpack_require__(/*! ./speecht5/feature_extraction_speecht5.js */ "./src/models/speecht5/feature_extraction_speecht5.js");
+/* harmony import */ var _wav2vec2_feature_extraction_wav2vec2_js__WEBPACK_IMPORTED_MODULE_6__ = __webpack_require__(/*! ./wav2vec2/feature_extraction_wav2vec2.js */ "./src/models/wav2vec2/feature_extraction_wav2vec2.js");
+/* harmony import */ var _wespeaker_feature_extraction_wespeaker_js__WEBPACK_IMPORTED_MODULE_7__ = __webpack_require__(/*! ./wespeaker/feature_extraction_wespeaker.js */ "./src/models/wespeaker/feature_extraction_wespeaker.js");
+/* harmony import */ var _whisper_feature_extraction_whisper_js__WEBPACK_IMPORTED_MODULE_8__ = __webpack_require__(/*! ./whisper/feature_extraction_whisper.js */ "./src/models/whisper/feature_extraction_whisper.js");
+/* harmony import */ var _base_image_processors_utils_js__WEBPACK_IMPORTED_MODULE_9__ = __webpack_require__(/*! ../base/image_processors_utils.js */ "./src/base/image_processors_utils.js");
+
 
 
 
@@ -17180,18 +17399,29 @@ class Idefics3ImageProcessor extends _base_image_processors_utils_js__WEBPACK_IM
             const optimal_width = Math.ceil(width / num_splits_w);
 
             // Iterate through each row and column
-            for (let r = 0; r < num_splits_h; r++) {
-                for (let c = 0; c < num_splits_w; c++) {
-                    // Calculate the starting point of the crop
-                    const start_x = c * optimal_width;
-                    const start_y = r * optimal_height;
-
-                    // Calculate the ending point of the crop
-                    const end_x = Math.min(start_x + optimal_width, width);
-                    const end_y = Math.min(start_y + optimal_height, height);
-
-                    // Crop the image
-                    frames.push(pixel_values.slice(null, null, [start_y, end_y], [start_x, end_x]));
+            for (let r = 0; r < num_splits_h; ++r) {
+                for (let c = 0; c < num_splits_w; ++c) {
+                    let start_x, start_y, end_x, end_y;
+                    if (r === num_splits_h - 1) { // At bottom
+                        start_y = height - optimal_height;
+                        end_y = height;
+                    } else {
+                        start_y = r * optimal_height;
+                        end_y = (r + 1) * optimal_height;
+                    }
+                    if (c === num_splits_w - 1) { // At right
+                        start_x = width - optimal_width;
+                        end_x = width;
+                    } else {
+                        start_x = c * optimal_width;
+                        end_x = (c + 1) * optimal_width;
+                    }
+
+                    const starts = [start_y, start_x];
+                    const ends = [end_y, end_x];
+
+                    const patch = await (0,_utils_tensor_js__WEBPACK_IMPORTED_MODULE_1__.slice)(pixel_values, starts, ends, [2, 3]);
+                    frames.push(patch);
                 }
             }
 
@@ -17417,21 +17647,22 @@ __webpack_require__.r(__webpack_exports__);
 /* harmony export */   OwlViTFeatureExtractor: () => (/* reexport safe */ _owlvit_image_processing_owlvit_js__WEBPACK_IMPORTED_MODULE_24__.OwlViTFeatureExtractor),
 /* harmony export */   OwlViTImageProcessor: () => (/* reexport safe */ _owlvit_image_processing_owlvit_js__WEBPACK_IMPORTED_MODULE_24__.OwlViTImageProcessor),
 /* harmony export */   Owlv2ImageProcessor: () => (/* reexport safe */ _owlv2_image_processing_owlv2_js__WEBPACK_IMPORTED_MODULE_23__.Owlv2ImageProcessor),
-/* harmony export */   PvtImageProcessor: () => (/* reexport safe */ _pvt_image_processing_pvt_js__WEBPACK_IMPORTED_MODULE_25__.PvtImageProcessor),
-/* harmony export */   Qwen2VLImageProcessor: () => (/* reexport safe */ _qwen2_vl_image_processing_qwen2_vl_js__WEBPACK_IMPORTED_MODULE_26__.Qwen2VLImageProcessor),
-/* harmony export */   RTDetrImageProcessor: () => (/* reexport safe */ _rt_detr_image_processing_rt_detr_js__WEBPACK_IMPORTED_MODULE_27__.RTDetrImageProcessor),
-/* harmony export */   SamImageProcessor: () => (/* reexport safe */ _sam_image_processing_sam_js__WEBPACK_IMPORTED_MODULE_28__.SamImageProcessor),
-/* harmony export */   SegformerFeatureExtractor: () => (/* reexport safe */ _segformer_image_processing_segformer_js__WEBPACK_IMPORTED_MODULE_29__.SegformerFeatureExtractor),
-/* harmony export */   SegformerImageProcessor: () => (/* reexport safe */ _segformer_image_processing_segformer_js__WEBPACK_IMPORTED_MODULE_29__.SegformerImageProcessor),
-/* harmony export */   SiglipImageProcessor: () => (/* reexport safe */ _siglip_image_processing_siglip_js__WEBPACK_IMPORTED_MODULE_30__.SiglipImageProcessor),
-/* harmony export */   Swin2SRImageProcessor: () => (/* reexport safe */ _swin2sr_image_processing_swin2sr_js__WEBPACK_IMPORTED_MODULE_31__.Swin2SRImageProcessor),
+/* harmony export */   Phi3VImageProcessor: () => (/* reexport safe */ _phi3_v_image_processing_phi3_v_js__WEBPACK_IMPORTED_MODULE_25__.Phi3VImageProcessor),
+/* harmony export */   PvtImageProcessor: () => (/* reexport safe */ _pvt_image_processing_pvt_js__WEBPACK_IMPORTED_MODULE_26__.PvtImageProcessor),
+/* harmony export */   Qwen2VLImageProcessor: () => (/* reexport safe */ _qwen2_vl_image_processing_qwen2_vl_js__WEBPACK_IMPORTED_MODULE_27__.Qwen2VLImageProcessor),
+/* harmony export */   RTDetrImageProcessor: () => (/* reexport safe */ _rt_detr_image_processing_rt_detr_js__WEBPACK_IMPORTED_MODULE_28__.RTDetrImageProcessor),
+/* harmony export */   SamImageProcessor: () => (/* reexport safe */ _sam_image_processing_sam_js__WEBPACK_IMPORTED_MODULE_29__.SamImageProcessor),
+/* harmony export */   SegformerFeatureExtractor: () => (/* reexport safe */ _segformer_image_processing_segformer_js__WEBPACK_IMPORTED_MODULE_30__.SegformerFeatureExtractor),
+/* harmony export */   SegformerImageProcessor: () => (/* reexport safe */ _segformer_image_processing_segformer_js__WEBPACK_IMPORTED_MODULE_30__.SegformerImageProcessor),
+/* harmony export */   SiglipImageProcessor: () => (/* reexport safe */ _siglip_image_processing_siglip_js__WEBPACK_IMPORTED_MODULE_31__.SiglipImageProcessor),
+/* harmony export */   Swin2SRImageProcessor: () => (/* reexport safe */ _swin2sr_image_processing_swin2sr_js__WEBPACK_IMPORTED_MODULE_32__.Swin2SRImageProcessor),
 /* harmony export */   VLMImageProcessor: () => (/* reexport safe */ _janus_image_processing_janus_js__WEBPACK_IMPORTED_MODULE_12__.VLMImageProcessor),
-/* harmony export */   ViTFeatureExtractor: () => (/* reexport safe */ _vit_image_processing_vit_js__WEBPACK_IMPORTED_MODULE_32__.ViTFeatureExtractor),
-/* harmony export */   ViTImageProcessor: () => (/* reexport safe */ _vit_image_processing_vit_js__WEBPACK_IMPORTED_MODULE_32__.ViTImageProcessor),
-/* harmony export */   VitMatteImageProcessor: () => (/* reexport safe */ _vitmatte_image_processing_vitmatte_js__WEBPACK_IMPORTED_MODULE_33__.VitMatteImageProcessor),
-/* harmony export */   VitPoseImageProcessor: () => (/* reexport safe */ _vitpose_image_processing_vitpose_js__WEBPACK_IMPORTED_MODULE_34__.VitPoseImageProcessor),
-/* harmony export */   YolosFeatureExtractor: () => (/* reexport safe */ _yolos_image_processing_yolos_js__WEBPACK_IMPORTED_MODULE_35__.YolosFeatureExtractor),
-/* harmony export */   YolosImageProcessor: () => (/* reexport safe */ _yolos_image_processing_yolos_js__WEBPACK_IMPORTED_MODULE_35__.YolosImageProcessor)
+/* harmony export */   ViTFeatureExtractor: () => (/* reexport safe */ _vit_image_processing_vit_js__WEBPACK_IMPORTED_MODULE_33__.ViTFeatureExtractor),
+/* harmony export */   ViTImageProcessor: () => (/* reexport safe */ _vit_image_processing_vit_js__WEBPACK_IMPORTED_MODULE_33__.ViTImageProcessor),
+/* harmony export */   VitMatteImageProcessor: () => (/* reexport safe */ _vitmatte_image_processing_vitmatte_js__WEBPACK_IMPORTED_MODULE_34__.VitMatteImageProcessor),
+/* harmony export */   VitPoseImageProcessor: () => (/* reexport safe */ _vitpose_image_processing_vitpose_js__WEBPACK_IMPORTED_MODULE_35__.VitPoseImageProcessor),
+/* harmony export */   YolosFeatureExtractor: () => (/* reexport safe */ _yolos_image_processing_yolos_js__WEBPACK_IMPORTED_MODULE_36__.YolosFeatureExtractor),
+/* harmony export */   YolosImageProcessor: () => (/* reexport safe */ _yolos_image_processing_yolos_js__WEBPACK_IMPORTED_MODULE_36__.YolosImageProcessor)
 /* harmony export */ });
 /* harmony import */ var _beit_image_processing_beit_js__WEBPACK_IMPORTED_MODULE_0__ = __webpack_require__(/*! ./beit/image_processing_beit.js */ "./src/models/beit/image_processing_beit.js");
 /* harmony import */ var _bit_image_processing_bit_js__WEBPACK_IMPORTED_MODULE_1__ = __webpack_require__(/*! ./bit/image_processing_bit.js */ "./src/models/bit/image_processing_bit.js");
@@ -17458,17 +17689,19 @@ __webpack_require__.r(__webpack_exports__);
 /* harmony import */ var _nougat_image_processing_nougat_js__WEBPACK_IMPORTED_MODULE_22__ = __webpack_require__(/*! ./nougat/image_processing_nougat.js */ "./src/models/nougat/image_processing_nougat.js");
 /* harmony import */ var _owlv2_image_processing_owlv2_js__WEBPACK_IMPORTED_MODULE_23__ = __webpack_require__(/*! ./owlv2/image_processing_owlv2.js */ "./src/models/owlv2/image_processing_owlv2.js");
 /* harmony import */ var _owlvit_image_processing_owlvit_js__WEBPACK_IMPORTED_MODULE_24__ = __webpack_require__(/*! ./owlvit/image_processing_owlvit.js */ "./src/models/owlvit/image_processing_owlvit.js");
-/* harmony import */ var _pvt_image_processing_pvt_js__WEBPACK_IMPORTED_MODULE_25__ = __webpack_require__(/*! ./pvt/image_processing_pvt.js */ "./src/models/pvt/image_processing_pvt.js");
-/* harmony import */ var _qwen2_vl_image_processing_qwen2_vl_js__WEBPACK_IMPORTED_MODULE_26__ = __webpack_require__(/*! ./qwen2_vl/image_processing_qwen2_vl.js */ "./src/models/qwen2_vl/image_processing_qwen2_vl.js");
-/* harmony import */ var _rt_detr_image_processing_rt_detr_js__WEBPACK_IMPORTED_MODULE_27__ = __webpack_require__(/*! ./rt_detr/image_processing_rt_detr.js */ "./src/models/rt_detr/image_processing_rt_detr.js");
-/* harmony import */ var _sam_image_processing_sam_js__WEBPACK_IMPORTED_MODULE_28__ = __webpack_require__(/*! ./sam/image_processing_sam.js */ "./src/models/sam/image_processing_sam.js");
-/* harmony import */ var _segformer_image_processing_segformer_js__WEBPACK_IMPORTED_MODULE_29__ = __webpack_require__(/*! ./segformer/image_processing_segformer.js */ "./src/models/segformer/image_processing_segformer.js");
-/* harmony import */ var _siglip_image_processing_siglip_js__WEBPACK_IMPORTED_MODULE_30__ = __webpack_require__(/*! ./siglip/image_processing_siglip.js */ "./src/models/siglip/image_processing_siglip.js");
-/* harmony import */ var _swin2sr_image_processing_swin2sr_js__WEBPACK_IMPORTED_MODULE_31__ = __webpack_require__(/*! ./swin2sr/image_processing_swin2sr.js */ "./src/models/swin2sr/image_processing_swin2sr.js");
-/* harmony import */ var _vit_image_processing_vit_js__WEBPACK_IMPORTED_MODULE_32__ = __webpack_require__(/*! ./vit/image_processing_vit.js */ "./src/models/vit/image_processing_vit.js");
-/* harmony import */ var _vitmatte_image_processing_vitmatte_js__WEBPACK_IMPORTED_MODULE_33__ = __webpack_require__(/*! ./vitmatte/image_processing_vitmatte.js */ "./src/models/vitmatte/image_processing_vitmatte.js");
-/* harmony import */ var _vitpose_image_processing_vitpose_js__WEBPACK_IMPORTED_MODULE_34__ = __webpack_require__(/*! ./vitpose/image_processing_vitpose.js */ "./src/models/vitpose/image_processing_vitpose.js");
-/* harmony import */ var _yolos_image_processing_yolos_js__WEBPACK_IMPORTED_MODULE_35__ = __webpack_require__(/*! ./yolos/image_processing_yolos.js */ "./src/models/yolos/image_processing_yolos.js");
+/* harmony import */ var _phi3_v_image_processing_phi3_v_js__WEBPACK_IMPORTED_MODULE_25__ = __webpack_require__(/*! ./phi3_v/image_processing_phi3_v.js */ "./src/models/phi3_v/image_processing_phi3_v.js");
+/* harmony import */ var _pvt_image_processing_pvt_js__WEBPACK_IMPORTED_MODULE_26__ = __webpack_require__(/*! ./pvt/image_processing_pvt.js */ "./src/models/pvt/image_processing_pvt.js");
+/* harmony import */ var _qwen2_vl_image_processing_qwen2_vl_js__WEBPACK_IMPORTED_MODULE_27__ = __webpack_require__(/*! ./qwen2_vl/image_processing_qwen2_vl.js */ "./src/models/qwen2_vl/image_processing_qwen2_vl.js");
+/* harmony import */ var _rt_detr_image_processing_rt_detr_js__WEBPACK_IMPORTED_MODULE_28__ = __webpack_require__(/*! ./rt_detr/image_processing_rt_detr.js */ "./src/models/rt_detr/image_processing_rt_detr.js");
+/* harmony import */ var _sam_image_processing_sam_js__WEBPACK_IMPORTED_MODULE_29__ = __webpack_require__(/*! ./sam/image_processing_sam.js */ "./src/models/sam/image_processing_sam.js");
+/* harmony import */ var _segformer_image_processing_segformer_js__WEBPACK_IMPORTED_MODULE_30__ = __webpack_require__(/*! ./segformer/image_processing_segformer.js */ "./src/models/segformer/image_processing_segformer.js");
+/* harmony import */ var _siglip_image_processing_siglip_js__WEBPACK_IMPORTED_MODULE_31__ = __webpack_require__(/*! ./siglip/image_processing_siglip.js */ "./src/models/siglip/image_processing_siglip.js");
+/* harmony import */ var _swin2sr_image_processing_swin2sr_js__WEBPACK_IMPORTED_MODULE_32__ = __webpack_require__(/*! ./swin2sr/image_processing_swin2sr.js */ "./src/models/swin2sr/image_processing_swin2sr.js");
+/* harmony import */ var _vit_image_processing_vit_js__WEBPACK_IMPORTED_MODULE_33__ = __webpack_require__(/*! ./vit/image_processing_vit.js */ "./src/models/vit/image_processing_vit.js");
+/* harmony import */ var _vitmatte_image_processing_vitmatte_js__WEBPACK_IMPORTED_MODULE_34__ = __webpack_require__(/*! ./vitmatte/image_processing_vitmatte.js */ "./src/models/vitmatte/image_processing_vitmatte.js");
+/* harmony import */ var _vitpose_image_processing_vitpose_js__WEBPACK_IMPORTED_MODULE_35__ = __webpack_require__(/*! ./vitpose/image_processing_vitpose.js */ "./src/models/vitpose/image_processing_vitpose.js");
+/* harmony import */ var _yolos_image_processing_yolos_js__WEBPACK_IMPORTED_MODULE_36__ = __webpack_require__(/*! ./yolos/image_processing_yolos.js */ "./src/models/yolos/image_processing_yolos.js");
+
 
 
 
@@ -18143,6 +18376,87 @@ class MobileViTImageProcessor extends _base_image_processors_utils_js__WEBPACK_I
 class MobileViTFeatureExtractor extends MobileViTImageProcessor { }
 
 
+/***/ }),
+
+/***/ "./src/models/moonshine/feature_extraction_moonshine.js":
+/*!**************************************************************!*\
+  !*** ./src/models/moonshine/feature_extraction_moonshine.js ***!
+  \**************************************************************/
+/***/ ((__unused_webpack___webpack_module__, __webpack_exports__, __webpack_require__) => {
+
+"use strict";
+__webpack_require__.r(__webpack_exports__);
+/* harmony export */ __webpack_require__.d(__webpack_exports__, {
+/* harmony export */   MoonshineFeatureExtractor: () => (/* binding */ MoonshineFeatureExtractor)
+/* harmony export */ });
+/* harmony import */ var _base_feature_extraction_utils_js__WEBPACK_IMPORTED_MODULE_0__ = __webpack_require__(/*! ../../base/feature_extraction_utils.js */ "./src/base/feature_extraction_utils.js");
+/* harmony import */ var _utils_tensor_js__WEBPACK_IMPORTED_MODULE_1__ = __webpack_require__(/*! ../../utils/tensor.js */ "./src/utils/tensor.js");
+
+
+
+
+class MoonshineFeatureExtractor extends _base_feature_extraction_utils_js__WEBPACK_IMPORTED_MODULE_0__.FeatureExtractor {
+    /**
+     * Asynchronously extracts input values from a given audio using the provided configuration.
+     * @param {Float32Array|Float64Array} audio The audio data as a Float32Array/Float64Array.
+     * @returns {Promise<{ input_values: Tensor; }>} The extracted input values.
+     */
+    async _call(audio) {
+        (0,_base_feature_extraction_utils_js__WEBPACK_IMPORTED_MODULE_0__.validate_audio_inputs)(audio, 'MoonshineFeatureExtractor');
+
+        if (audio instanceof Float64Array) {
+            audio = new Float32Array(audio);
+        }
+
+        const shape = [
+            1,            /* batch_size */
+            audio.length, /* num_samples */
+        ];
+        return {
+            input_values: new _utils_tensor_js__WEBPACK_IMPORTED_MODULE_1__.Tensor('float32', audio, shape),
+        };
+    }
+}
+
+
+/***/ }),
+
+/***/ "./src/models/moonshine/processing_moonshine.js":
+/*!******************************************************!*\
+  !*** ./src/models/moonshine/processing_moonshine.js ***!
+  \******************************************************/
+/***/ ((__unused_webpack___webpack_module__, __webpack_exports__, __webpack_require__) => {
+
+"use strict";
+__webpack_require__.r(__webpack_exports__);
+/* harmony export */ __webpack_require__.d(__webpack_exports__, {
+/* harmony export */   MoonshineProcessor: () => (/* binding */ MoonshineProcessor)
+/* harmony export */ });
+/* harmony import */ var _auto_feature_extraction_auto_js__WEBPACK_IMPORTED_MODULE_0__ = __webpack_require__(/*! ../auto/feature_extraction_auto.js */ "./src/models/auto/feature_extraction_auto.js");
+/* harmony import */ var _tokenizers_js__WEBPACK_IMPORTED_MODULE_1__ = __webpack_require__(/*! ../../tokenizers.js */ "./src/tokenizers.js");
+/* harmony import */ var _base_processing_utils_js__WEBPACK_IMPORTED_MODULE_2__ = __webpack_require__(/*! ../../base/processing_utils.js */ "./src/base/processing_utils.js");
+
+
+
+
+/**
+ * Represents a MoonshineProcessor that extracts features from an audio input.
+ */
+class MoonshineProcessor extends _base_processing_utils_js__WEBPACK_IMPORTED_MODULE_2__.Processor {
+    static tokenizer_class = _tokenizers_js__WEBPACK_IMPORTED_MODULE_1__.AutoTokenizer
+    static feature_extractor_class = _auto_feature_extraction_auto_js__WEBPACK_IMPORTED_MODULE_0__.AutoFeatureExtractor
+
+    /**
+     * Calls the feature_extractor function with the given audio input.
+     * @param {any} audio The audio input to extract features from.
+     * @returns {Promise<any>} A Promise that resolves with the extracted features.
+     */
+    async _call(audio) {
+        return await this.feature_extractor(audio);
+    }
+}
+
+
 /***/ }),
 
 /***/ "./src/models/nougat/image_processing_nougat.js":
@@ -18336,6 +18650,256 @@ class PaliGemmaProcessor extends _base_processing_utils_js__WEBPACK_IMPORTED_MOD
 }
 
 
+/***/ }),
+
+/***/ "./src/models/phi3_v/image_processing_phi3_v.js":
+/*!******************************************************!*\
+  !*** ./src/models/phi3_v/image_processing_phi3_v.js ***!
+  \******************************************************/
+/***/ ((__unused_webpack___webpack_module__, __webpack_exports__, __webpack_require__) => {
+
+"use strict";
+__webpack_require__.r(__webpack_exports__);
+/* harmony export */ __webpack_require__.d(__webpack_exports__, {
+/* harmony export */   Phi3VImageProcessor: () => (/* binding */ Phi3VImageProcessor)
+/* harmony export */ });
+/* harmony import */ var _base_image_processors_utils_js__WEBPACK_IMPORTED_MODULE_0__ = __webpack_require__(/*! ../../base/image_processors_utils.js */ "./src/base/image_processors_utils.js");
+/* harmony import */ var _utils_tensor_js__WEBPACK_IMPORTED_MODULE_1__ = __webpack_require__(/*! ../../utils/tensor.js */ "./src/utils/tensor.js");
+
+
+
+const IMAGE_SIZE = 336;
+const SLICE_AXES = [2, 3]; // axes to slice on
+const { ceil, floor, sqrt } = Math;
+
+class Phi3VImageProcessor extends _base_image_processors_utils_js__WEBPACK_IMPORTED_MODULE_0__.ImageProcessor {
+    constructor(config) {
+        super({
+            ...config,
+            do_normalize: true,
+            do_pad: true,
+            pad_size: 'custom',
+            do_convert_rgb: true,
+            do_resize: true, // Smart resizing "hd_transform"
+        });
+
+        this._num_crops = config.num_crops;
+    }
+    calc_num_image_tokens_from_image_size(width, height) {
+        // @ts-expect-error
+        const { num_img_tokens } = this.config;
+        return floor(((floor((height / IMAGE_SIZE)) * floor((width / IMAGE_SIZE)) + 1) * num_img_tokens) + 1 + (floor(height / IMAGE_SIZE) + 1) * sqrt(num_img_tokens));
+    }
+
+    /** @type {ImageProcessor['get_resize_output_image_size']} */
+    get_resize_output_image_size(image, size) {
+        const hd_num = this._num_crops;
+        const [width, height] = image.size
+
+        let ratio = width / height;
+        let scale = 1;
+
+        // Calculate the scaling factor
+        while (scale * Math.ceil(scale / ratio) <= hd_num) {
+            scale += 1;
+        }
+        scale -= 1;
+
+        // Compute the new dimensions
+        const new_w = Math.floor(scale * 336);
+        const new_h = Math.floor(new_w / ratio);
+
+        return [new_w, new_h]
+    }
+
+
+    /** @type {ImageProcessor['pad_image']} */
+    pad_image(pixelData, imgDims, padSize, options = {}) {
+        // Phi3V uses a custom padding strategy:
+        // - Pad to a multiple of 336
+        // - Pad with white pixels
+        const [imageHeight, imageWidth] = imgDims;
+        const height = IMAGE_SIZE * ceil(imageHeight / IMAGE_SIZE);
+        const width = IMAGE_SIZE * ceil(imageWidth / IMAGE_SIZE);
+
+        // NOTE: Since padding is done after normalization, we need to fill with the normalized values
+        const constant_values = [1, 1, 1].map((x, i) => (x - this.image_mean[i]) / this.image_std[i]);
+        return super.pad_image(pixelData, imgDims, { width, height }, {
+            center: true,
+            constant_values,
+            ...options,
+        });
+    }
+
+    async _call(images, {
+        num_crops = null,
+    } = {}) {
+        // @ts-expect-error
+        this._num_crops = num_crops ??= this.config.num_crops;
+        if (num_crops < 4 || sqrt(num_crops) % 1 !== 0) {
+            throw new Error("num_crops must be a square number >= 4");
+        }
+
+        if (!Array.isArray(images)) {
+            images = [images];
+        }
+
+        const num_images = images.length;
+        const imageData = await Promise.all(images.map(x => this.preprocess(x)));
+
+        const original_sizes = imageData.map(x => x.original_size);
+        const reshaped_input_sizes = imageData.map(x => x.reshaped_input_size);
+
+        // Process each image in batch
+        const all_pixel_values = [];
+        for (const { pixel_values } of imageData) {
+            pixel_values.unsqueeze_(0); // Easier processing as 4D tensor
+
+            const [height, width] = pixel_values.dims.slice(-2);
+
+            // Global image (Tensor of shape [num_channels, height, width])
+            const batch_pixel_values = await (0,_utils_tensor_js__WEBPACK_IMPORTED_MODULE_1__.interpolate_4d)(pixel_values, {
+                size: [IMAGE_SIZE, IMAGE_SIZE],
+                mode: 'bicubic',
+            });
+
+            if (num_crops > 0) {
+                const patches = [];
+                const sqrt_patches = sqrt(num_crops);
+                const patch_width = floor(width / sqrt_patches);
+                const patch_height = floor(height / sqrt_patches);
+                for (let y = 0; y < sqrt_patches; ++y) {
+                    for (let x = 0; x < sqrt_patches; ++x) {
+                        let start_x, start_y, end_x, end_y;
+                        if (y === sqrt_patches - 1) { // At bottom
+                            start_y = height - patch_height;
+                            end_y = height;
+                        } else {
+                            start_y = y * patch_height;
+                            end_y = (y + 1) * patch_height;
+                        }
+                        if (x === sqrt_patches - 1) { // At right
+                            start_x = width - patch_width;
+                            end_x = width;
+                        } else {
+                            start_x = x * patch_width;
+                            end_x = (x + 1) * patch_width;
+                        }
+
+                        const starts = [start_y, start_x];
+                        const ends = [end_y, end_x];
+                        const patch = await (0,_utils_tensor_js__WEBPACK_IMPORTED_MODULE_1__.slice)(pixel_values, starts, ends, SLICE_AXES);
+                        patches.push(patch);
+                    }
+                }
+
+                const resized_tensors = await (0,_utils_tensor_js__WEBPACK_IMPORTED_MODULE_1__.interpolate_4d)((0,_utils_tensor_js__WEBPACK_IMPORTED_MODULE_1__.cat)(patches, 0), {
+                    size: [IMAGE_SIZE, IMAGE_SIZE],
+                    mode: 'bicubic',
+                }); // [num_crops, 3, 336, 336]
+
+                // Concatenate the global image with the patches
+                all_pixel_values.push((0,_utils_tensor_js__WEBPACK_IMPORTED_MODULE_1__.cat)([batch_pixel_values, resized_tensors], 0));
+            } else {
+                // Only use the global image
+                // NOTE: Not currently supported in modelling code
+                all_pixel_values.push(batch_pixel_values);
+            }
+        }
+
+        // [num_images, 1 + num_crops, num_channels=3, height, width]
+        const pixel_values = (0,_utils_tensor_js__WEBPACK_IMPORTED_MODULE_1__.stack)(all_pixel_values, 0);
+
+        // Calculate padded image sizes
+        const sizes = reshaped_input_sizes.map(x => x.map(y => IMAGE_SIZE * ceil(y / IMAGE_SIZE)));
+
+        const image_sizes = new _utils_tensor_js__WEBPACK_IMPORTED_MODULE_1__.Tensor(
+            'int64',
+            sizes.flat(),
+            [num_images, 2],
+        );
+
+        const num_img_tokens = sizes.map(
+            ([height, width]) => this.calc_num_image_tokens_from_image_size(width, height),
+        );
+
+        return { pixel_values, original_sizes, reshaped_input_sizes, image_sizes, num_img_tokens };
+    }
+}
+
+
+/***/ }),
+
+/***/ "./src/models/phi3_v/processing_phi3_v.js":
+/*!************************************************!*\
+  !*** ./src/models/phi3_v/processing_phi3_v.js ***!
+  \************************************************/
+/***/ ((__unused_webpack___webpack_module__, __webpack_exports__, __webpack_require__) => {
+
+"use strict";
+__webpack_require__.r(__webpack_exports__);
+/* harmony export */ __webpack_require__.d(__webpack_exports__, {
+/* harmony export */   Phi3VProcessor: () => (/* binding */ Phi3VProcessor)
+/* harmony export */ });
+/* harmony import */ var _base_processing_utils_js__WEBPACK_IMPORTED_MODULE_0__ = __webpack_require__(/*! ../../base/processing_utils.js */ "./src/base/processing_utils.js");
+/* harmony import */ var _auto_image_processing_auto_js__WEBPACK_IMPORTED_MODULE_1__ = __webpack_require__(/*! ../auto/image_processing_auto.js */ "./src/models/auto/image_processing_auto.js");
+/* harmony import */ var _tokenizers_js__WEBPACK_IMPORTED_MODULE_2__ = __webpack_require__(/*! ../../tokenizers.js */ "./src/tokenizers.js");
+/* harmony import */ var _utils_image_js__WEBPACK_IMPORTED_MODULE_3__ = __webpack_require__(/*! ../../utils/image.js */ "./src/utils/image.js");
+
+
+
+
+
+const IMAGE_TOKEN = "<|image|>";
+const IMAGE_TOKEN_PATTERN = /<\|image_\d+\|>/g;
+
+class Phi3VProcessor extends _base_processing_utils_js__WEBPACK_IMPORTED_MODULE_0__.Processor {
+    static image_processor_class = _auto_image_processing_auto_js__WEBPACK_IMPORTED_MODULE_1__.AutoImageProcessor
+    static tokenizer_class = _tokenizers_js__WEBPACK_IMPORTED_MODULE_2__.AutoTokenizer
+
+    /**
+     * 
+     * @param {string|string[]} text 
+     * @param {RawImage|RawImage[]} images 
+     * @param  {...any} args 
+     * @returns {Promise<any>}
+     */
+    async _call(text, images = null, {
+        padding = true,
+        truncation = true,
+        num_crops = null,
+    } = {}) {
+
+        if (!Array.isArray(text)) {
+            text = [text];
+        }
+
+        let text_inputs, image_inputs;
+        if (images) {
+            image_inputs = await this.image_processor(images, { num_crops });
+            const { num_img_tokens } = image_inputs;
+
+            // The original implementation adds a bos_token before the image tokens
+            // TODO: Check if this affects performance, since it looks like a bug in the original implementation
+            const prompt_chunks = text.map((t, i) => t.split(IMAGE_TOKEN_PATTERN).join(IMAGE_TOKEN.repeat(num_img_tokens[i])));
+
+            text_inputs = this.tokenizer(prompt_chunks, { padding, truncation });
+
+            // The model expects image tokens to be negative, so we negate the image token ids
+            const image_token_id = this.tokenizer.model.convert_tokens_to_ids([IMAGE_TOKEN])[0];
+            text_inputs.input_ids.map_(id => (id == image_token_id) ? -id : id);
+        } else {
+            text_inputs = this.tokenizer(text);
+        }
+
+        return {
+            ...text_inputs,
+            ...image_inputs,
+        }
+    }
+}
+
+
 /***/ }),
 
 /***/ "./src/models/processors.js":
@@ -18348,32 +18912,38 @@ class PaliGemmaProcessor extends _base_processing_utils_js__WEBPACK_IMPORTED_MOD
 __webpack_require__.r(__webpack_exports__);
 /* harmony export */ __webpack_require__.d(__webpack_exports__, {
 /* harmony export */   Florence2Processor: () => (/* reexport safe */ _florence2_processing_florence2_js__WEBPACK_IMPORTED_MODULE_0__.Florence2Processor),
-/* harmony export */   Idefics3Processor: () => (/* reexport safe */ _idefics3_processing_idefics3_js__WEBPACK_IMPORTED_MODULE_2__.Idefics3Processor),
-/* harmony export */   JinaCLIPProcessor: () => (/* reexport safe */ _jina_clip_processing_jina_clip_js__WEBPACK_IMPORTED_MODULE_4__.JinaCLIPProcessor),
+/* harmony export */   Idefics3Processor: () => (/* reexport safe */ _idefics3_processing_idefics3_js__WEBPACK_IMPORTED_MODULE_3__.Idefics3Processor),
+/* harmony export */   JinaCLIPProcessor: () => (/* reexport safe */ _jina_clip_processing_jina_clip_js__WEBPACK_IMPORTED_MODULE_5__.JinaCLIPProcessor),
 /* harmony export */   MgpstrProcessor: () => (/* reexport safe */ _mgp_str_processing_mgp_str_js__WEBPACK_IMPORTED_MODULE_1__.MgpstrProcessor),
-/* harmony export */   OwlViTProcessor: () => (/* reexport safe */ _owlvit_processing_owlvit_js__WEBPACK_IMPORTED_MODULE_5__.OwlViTProcessor),
-/* harmony export */   PaliGemmaProcessor: () => (/* reexport safe */ _paligemma_processing_paligemma_js__WEBPACK_IMPORTED_MODULE_6__.PaliGemmaProcessor),
-/* harmony export */   PyAnnoteProcessor: () => (/* reexport safe */ _pyannote_processing_pyannote_js__WEBPACK_IMPORTED_MODULE_7__.PyAnnoteProcessor),
-/* harmony export */   Qwen2VLProcessor: () => (/* reexport safe */ _qwen2_vl_processing_qwen2_vl_js__WEBPACK_IMPORTED_MODULE_8__.Qwen2VLProcessor),
-/* harmony export */   SamProcessor: () => (/* reexport safe */ _sam_processing_sam_js__WEBPACK_IMPORTED_MODULE_9__.SamProcessor),
-/* harmony export */   SpeechT5Processor: () => (/* reexport safe */ _speecht5_processing_speecht5_js__WEBPACK_IMPORTED_MODULE_10__.SpeechT5Processor),
-/* harmony export */   VLChatProcessor: () => (/* reexport safe */ _janus_processing_janus_js__WEBPACK_IMPORTED_MODULE_3__.VLChatProcessor),
-/* harmony export */   Wav2Vec2ProcessorWithLM: () => (/* reexport safe */ _wav2vec2_processing_wav2vec2_js__WEBPACK_IMPORTED_MODULE_11__.Wav2Vec2ProcessorWithLM),
-/* harmony export */   WhisperProcessor: () => (/* reexport safe */ _whisper_processing_whisper_js__WEBPACK_IMPORTED_MODULE_12__.WhisperProcessor)
+/* harmony export */   MoonshineProcessor: () => (/* reexport safe */ _moonshine_processing_moonshine_js__WEBPACK_IMPORTED_MODULE_2__.MoonshineProcessor),
+/* harmony export */   OwlViTProcessor: () => (/* reexport safe */ _owlvit_processing_owlvit_js__WEBPACK_IMPORTED_MODULE_6__.OwlViTProcessor),
+/* harmony export */   PaliGemmaProcessor: () => (/* reexport safe */ _paligemma_processing_paligemma_js__WEBPACK_IMPORTED_MODULE_8__.PaliGemmaProcessor),
+/* harmony export */   Phi3VProcessor: () => (/* reexport safe */ _phi3_v_processing_phi3_v_js__WEBPACK_IMPORTED_MODULE_7__.Phi3VProcessor),
+/* harmony export */   PyAnnoteProcessor: () => (/* reexport safe */ _pyannote_processing_pyannote_js__WEBPACK_IMPORTED_MODULE_9__.PyAnnoteProcessor),
+/* harmony export */   Qwen2VLProcessor: () => (/* reexport safe */ _qwen2_vl_processing_qwen2_vl_js__WEBPACK_IMPORTED_MODULE_10__.Qwen2VLProcessor),
+/* harmony export */   SamProcessor: () => (/* reexport safe */ _sam_processing_sam_js__WEBPACK_IMPORTED_MODULE_11__.SamProcessor),
+/* harmony export */   SpeechT5Processor: () => (/* reexport safe */ _speecht5_processing_speecht5_js__WEBPACK_IMPORTED_MODULE_12__.SpeechT5Processor),
+/* harmony export */   VLChatProcessor: () => (/* reexport safe */ _janus_processing_janus_js__WEBPACK_IMPORTED_MODULE_4__.VLChatProcessor),
+/* harmony export */   Wav2Vec2ProcessorWithLM: () => (/* reexport safe */ _wav2vec2_processing_wav2vec2_js__WEBPACK_IMPORTED_MODULE_13__.Wav2Vec2ProcessorWithLM),
+/* harmony export */   WhisperProcessor: () => (/* reexport safe */ _whisper_processing_whisper_js__WEBPACK_IMPORTED_MODULE_14__.WhisperProcessor)
 /* harmony export */ });
 /* harmony import */ var _florence2_processing_florence2_js__WEBPACK_IMPORTED_MODULE_0__ = __webpack_require__(/*! ./florence2/processing_florence2.js */ "./src/models/florence2/processing_florence2.js");
 /* harmony import */ var _mgp_str_processing_mgp_str_js__WEBPACK_IMPORTED_MODULE_1__ = __webpack_require__(/*! ./mgp_str/processing_mgp_str.js */ "./src/models/mgp_str/processing_mgp_str.js");
-/* harmony import */ var _idefics3_processing_idefics3_js__WEBPACK_IMPORTED_MODULE_2__ = __webpack_require__(/*! ./idefics3/processing_idefics3.js */ "./src/models/idefics3/processing_idefics3.js");
-/* harmony import */ var _janus_processing_janus_js__WEBPACK_IMPORTED_MODULE_3__ = __webpack_require__(/*! ./janus/processing_janus.js */ "./src/models/janus/processing_janus.js");
-/* harmony import */ var _jina_clip_processing_jina_clip_js__WEBPACK_IMPORTED_MODULE_4__ = __webpack_require__(/*! ./jina_clip/processing_jina_clip.js */ "./src/models/jina_clip/processing_jina_clip.js");
-/* harmony import */ var _owlvit_processing_owlvit_js__WEBPACK_IMPORTED_MODULE_5__ = __webpack_require__(/*! ./owlvit/processing_owlvit.js */ "./src/models/owlvit/processing_owlvit.js");
-/* harmony import */ var _paligemma_processing_paligemma_js__WEBPACK_IMPORTED_MODULE_6__ = __webpack_require__(/*! ./paligemma/processing_paligemma.js */ "./src/models/paligemma/processing_paligemma.js");
-/* harmony import */ var _pyannote_processing_pyannote_js__WEBPACK_IMPORTED_MODULE_7__ = __webpack_require__(/*! ./pyannote/processing_pyannote.js */ "./src/models/pyannote/processing_pyannote.js");
-/* harmony import */ var _qwen2_vl_processing_qwen2_vl_js__WEBPACK_IMPORTED_MODULE_8__ = __webpack_require__(/*! ./qwen2_vl/processing_qwen2_vl.js */ "./src/models/qwen2_vl/processing_qwen2_vl.js");
-/* harmony import */ var _sam_processing_sam_js__WEBPACK_IMPORTED_MODULE_9__ = __webpack_require__(/*! ./sam/processing_sam.js */ "./src/models/sam/processing_sam.js");
-/* harmony import */ var _speecht5_processing_speecht5_js__WEBPACK_IMPORTED_MODULE_10__ = __webpack_require__(/*! ./speecht5/processing_speecht5.js */ "./src/models/speecht5/processing_speecht5.js");
-/* harmony import */ var _wav2vec2_processing_wav2vec2_js__WEBPACK_IMPORTED_MODULE_11__ = __webpack_require__(/*! ./wav2vec2/processing_wav2vec2.js */ "./src/models/wav2vec2/processing_wav2vec2.js");
-/* harmony import */ var _whisper_processing_whisper_js__WEBPACK_IMPORTED_MODULE_12__ = __webpack_require__(/*! ./whisper/processing_whisper.js */ "./src/models/whisper/processing_whisper.js");
+/* harmony import */ var _moonshine_processing_moonshine_js__WEBPACK_IMPORTED_MODULE_2__ = __webpack_require__(/*! ./moonshine/processing_moonshine.js */ "./src/models/moonshine/processing_moonshine.js");
+/* harmony import */ var _idefics3_processing_idefics3_js__WEBPACK_IMPORTED_MODULE_3__ = __webpack_require__(/*! ./idefics3/processing_idefics3.js */ "./src/models/idefics3/processing_idefics3.js");
+/* harmony import */ var _janus_processing_janus_js__WEBPACK_IMPORTED_MODULE_4__ = __webpack_require__(/*! ./janus/processing_janus.js */ "./src/models/janus/processing_janus.js");
+/* harmony import */ var _jina_clip_processing_jina_clip_js__WEBPACK_IMPORTED_MODULE_5__ = __webpack_require__(/*! ./jina_clip/processing_jina_clip.js */ "./src/models/jina_clip/processing_jina_clip.js");
+/* harmony import */ var _owlvit_processing_owlvit_js__WEBPACK_IMPORTED_MODULE_6__ = __webpack_require__(/*! ./owlvit/processing_owlvit.js */ "./src/models/owlvit/processing_owlvit.js");
+/* harmony import */ var _phi3_v_processing_phi3_v_js__WEBPACK_IMPORTED_MODULE_7__ = __webpack_require__(/*! ./phi3_v/processing_phi3_v.js */ "./src/models/phi3_v/processing_phi3_v.js");
+/* harmony import */ var _paligemma_processing_paligemma_js__WEBPACK_IMPORTED_MODULE_8__ = __webpack_require__(/*! ./paligemma/processing_paligemma.js */ "./src/models/paligemma/processing_paligemma.js");
+/* harmony import */ var _pyannote_processing_pyannote_js__WEBPACK_IMPORTED_MODULE_9__ = __webpack_require__(/*! ./pyannote/processing_pyannote.js */ "./src/models/pyannote/processing_pyannote.js");
+/* harmony import */ var _qwen2_vl_processing_qwen2_vl_js__WEBPACK_IMPORTED_MODULE_10__ = __webpack_require__(/*! ./qwen2_vl/processing_qwen2_vl.js */ "./src/models/qwen2_vl/processing_qwen2_vl.js");
+/* harmony import */ var _sam_processing_sam_js__WEBPACK_IMPORTED_MODULE_11__ = __webpack_require__(/*! ./sam/processing_sam.js */ "./src/models/sam/processing_sam.js");
+/* harmony import */ var _speecht5_processing_speecht5_js__WEBPACK_IMPORTED_MODULE_12__ = __webpack_require__(/*! ./speecht5/processing_speecht5.js */ "./src/models/speecht5/processing_speecht5.js");
+/* harmony import */ var _wav2vec2_processing_wav2vec2_js__WEBPACK_IMPORTED_MODULE_13__ = __webpack_require__(/*! ./wav2vec2/processing_wav2vec2.js */ "./src/models/wav2vec2/processing_wav2vec2.js");
+/* harmony import */ var _whisper_processing_whisper_js__WEBPACK_IMPORTED_MODULE_14__ = __webpack_require__(/*! ./whisper/processing_whisper.js */ "./src/models/whisper/processing_whisper.js");
+
+
 
 
 
@@ -18423,6 +18993,8 @@ __webpack_require__.r(__webpack_exports__);
 /* harmony export */ });
 /* harmony import */ var _base_feature_extraction_utils_js__WEBPACK_IMPORTED_MODULE_0__ = __webpack_require__(/*! ../../base/feature_extraction_utils.js */ "./src/base/feature_extraction_utils.js");
 /* harmony import */ var _utils_tensor_js__WEBPACK_IMPORTED_MODULE_1__ = __webpack_require__(/*! ../../utils/tensor.js */ "./src/utils/tensor.js");
+/* harmony import */ var _utils_maths_js__WEBPACK_IMPORTED_MODULE_2__ = __webpack_require__(/*! ../../utils/maths.js */ "./src/utils/maths.js");
+
 
 
 
@@ -18450,41 +19022,6 @@ class PyAnnoteFeatureExtractor extends _base_feature_extraction_utils_js__WEBPAC
         };
     }
 
-}
-
-
-/***/ }),
-
-/***/ "./src/models/pyannote/processing_pyannote.js":
-/*!****************************************************!*\
-  !*** ./src/models/pyannote/processing_pyannote.js ***!
-  \****************************************************/
-/***/ ((__unused_webpack___webpack_module__, __webpack_exports__, __webpack_require__) => {
-
-"use strict";
-__webpack_require__.r(__webpack_exports__);
-/* harmony export */ __webpack_require__.d(__webpack_exports__, {
-/* harmony export */   PyAnnoteProcessor: () => (/* binding */ PyAnnoteProcessor)
-/* harmony export */ });
-/* harmony import */ var _base_processing_utils_js__WEBPACK_IMPORTED_MODULE_0__ = __webpack_require__(/*! ../../base/processing_utils.js */ "./src/base/processing_utils.js");
-/* harmony import */ var _auto_feature_extraction_auto_js__WEBPACK_IMPORTED_MODULE_1__ = __webpack_require__(/*! ../auto/feature_extraction_auto.js */ "./src/models/auto/feature_extraction_auto.js");
-/* harmony import */ var _utils_maths_js__WEBPACK_IMPORTED_MODULE_2__ = __webpack_require__(/*! ../../utils/maths.js */ "./src/utils/maths.js");
-
-
-
-
-class PyAnnoteProcessor extends _base_processing_utils_js__WEBPACK_IMPORTED_MODULE_0__.Processor {
-    static feature_extractor_class = _auto_feature_extraction_auto_js__WEBPACK_IMPORTED_MODULE_1__.AutoFeatureExtractor
-
-    /**
-     * Calls the feature_extractor function with the given audio input.
-     * @param {any} audio The audio input to extract features from.
-     * @returns {Promise<any>} A Promise that resolves with the extracted features.
-     */
-    async _call(audio) {
-        return await this.feature_extractor(audio)
-    }
-
     /**
      * NOTE: Can return fractional values. `Math.ceil` will ensure correct value.
      * @param {number} samples The number of frames in the audio.
@@ -18539,6 +19076,48 @@ class PyAnnoteProcessor extends _base_processing_utils_js__WEBPACK_IMPORTED_MODU
         }
         return results;
     }
+
+}
+
+
+/***/ }),
+
+/***/ "./src/models/pyannote/processing_pyannote.js":
+/*!****************************************************!*\
+  !*** ./src/models/pyannote/processing_pyannote.js ***!
+  \****************************************************/
+/***/ ((__unused_webpack___webpack_module__, __webpack_exports__, __webpack_require__) => {
+
+"use strict";
+__webpack_require__.r(__webpack_exports__);
+/* harmony export */ __webpack_require__.d(__webpack_exports__, {
+/* harmony export */   PyAnnoteProcessor: () => (/* binding */ PyAnnoteProcessor)
+/* harmony export */ });
+/* harmony import */ var _base_processing_utils_js__WEBPACK_IMPORTED_MODULE_0__ = __webpack_require__(/*! ../../base/processing_utils.js */ "./src/base/processing_utils.js");
+/* harmony import */ var _feature_extraction_pyannote_js__WEBPACK_IMPORTED_MODULE_1__ = __webpack_require__(/*! ./feature_extraction_pyannote.js */ "./src/models/pyannote/feature_extraction_pyannote.js");
+
+
+
+class PyAnnoteProcessor extends _base_processing_utils_js__WEBPACK_IMPORTED_MODULE_0__.Processor {
+    static feature_extractor_class = _feature_extraction_pyannote_js__WEBPACK_IMPORTED_MODULE_1__.PyAnnoteFeatureExtractor
+
+    /**
+     * Calls the feature_extractor function with the given audio input.
+     * @param {any} audio The audio input to extract features from.
+     * @returns {Promise<any>} A Promise that resolves with the extracted features.
+     */
+    async _call(audio) {
+        return await this.feature_extractor(audio)
+    }
+
+    /** @type {PyAnnoteFeatureExtractor['post_process_speaker_diarization']} */
+    post_process_speaker_diarization(...args) {
+        return /** @type {PyAnnoteFeatureExtractor} */(this.feature_extractor).post_process_speaker_diarization(...args);
+    }
+
+    get sampling_rate() {
+        return this.feature_extractor.config.sampling_rate;
+    }
 }
 
 
@@ -20288,6 +20867,17 @@ class TensorOpRegistry {
         }
         return this._top_k;
     }
+
+    static get slice() {
+        if (!this._slice) {
+            this._slice = wrap(
+                [8, 7, 18, 0, 58, 96, 10, 25, 10, 1, 120, 10, 1, 115, 10, 1, 101, 10, 1, 97, 10, 1, 116, 18, 1, 121, 34, 5, 83, 108, 105, 99, 101, 18, 1, 114, 90, 9, 10, 1, 120, 18, 4, 10, 2, 8, 1, 90, 9, 10, 1, 115, 18, 4, 10, 2, 8, 7, 90, 9, 10, 1, 101, 18, 4, 10, 2, 8, 7, 90, 9, 10, 1, 97, 18, 4, 10, 2, 8, 7, 90, 9, 10, 1, 116, 18, 4, 10, 2, 8, 7, 98, 9, 10, 1, 121, 18, 4, 10, 2, 8, 1, 66, 2, 16, 13],
+                this.session_options,
+                'y',
+            )
+        }
+        return this._slice;
+    }
 }
 
 
@@ -20980,7 +21570,7 @@ class FillMaskPipeline extends (/** @type {new (options: TextPipelineConstructor
                 return {
                     score: values[i],
                     token: Number(x),
-                    token_str: this.tokenizer.model.vocab[x],
+                    token_str: this.tokenizer.decode([x]),
                     sequence: this.tokenizer.decode(sequence, { skip_special_tokens: true }),
                 }
             }));
@@ -22021,6 +22611,8 @@ class AutomaticSpeechRecognitionPipeline extends (/** @type {new (options: TextA
             case 'unispeech-sat':
             case 'hubert':
                 return this._call_wav2vec2(audio, kwargs)
+            case 'moonshine':
+                return this._call_moonshine(audio, kwargs)
             default:
                 throw new Error(`AutomaticSpeechRecognitionPipeline does not support model type '${this.model.config.model_type}'.`)
         }
@@ -22174,6 +22766,34 @@ class AutomaticSpeechRecognitionPipeline extends (/** @type {new (options: TextA
         }
         return single ? toReturn[0] : toReturn;
     }
+
+    /**
+     * @type {AutomaticSpeechRecognitionPipelineCallback}
+     * @private
+     */
+    async _call_moonshine(audio, kwargs) {
+        const single = !Array.isArray(audio);
+        if (single) {
+            audio = [/** @type {AudioInput} */ (audio)];
+        }
+        const sampling_rate = this.processor.feature_extractor.config.sampling_rate;
+        const preparedAudios = await prepareAudios(audio, sampling_rate);
+        const toReturn = [];
+        for (const aud of preparedAudios) {
+            const inputs = await this.processor(aud);
+
+            // According to the [paper](https://arxiv.org/pdf/2410.15608):
+            // "We use greedy decoding, with a heuristic limit of 6 output tokens
+            // per second of audio to avoid repeated output sequences."
+            const max_new_tokens = Math.floor(aud.length / sampling_rate) * 6;
+            const outputs = await this.model.generate({ max_new_tokens, ...kwargs, ...inputs });
+
+            const text = this.processor.batch_decode(outputs, { skip_special_tokens: true })[0];
+            toReturn.push({ text });
+        }
+        return single ? toReturn[0] : toReturn;
+    }
+
 }
 
 /**
@@ -32372,7 +32992,9 @@ __webpack_require__.r(__webpack_exports__);
 /* harmony export */   ones_like: () => (/* binding */ ones_like),
 /* harmony export */   permute: () => (/* binding */ permute),
 /* harmony export */   quantize_embeddings: () => (/* binding */ quantize_embeddings),
+/* harmony export */   rand: () => (/* binding */ rand),
 /* harmony export */   rfft: () => (/* binding */ rfft),
+/* harmony export */   slice: () => (/* binding */ slice),
 /* harmony export */   stack: () => (/* binding */ stack),
 /* harmony export */   std_mean: () => (/* binding */ std_mean),
 /* harmony export */   topk: () => (/* binding */ topk),
@@ -33151,8 +33773,21 @@ class Tensor {
         if (!DataTypeMap.hasOwnProperty(type)) {
             throw new Error(`Unsupported type: ${type}`);
         }
+
+        // Handle special cases where a mapping function is needed (e.g., where one type is a bigint and the other is a number)
+        let map_fn;
+        const is_source_bigint = ['int64', 'uint64'].includes(this.type);
+        const is_dest_bigint = ['int64', 'uint64'].includes(type);
+        if (is_source_bigint && !is_dest_bigint) {
+            // TypeError: Cannot convert a BigInt value to a number
+            map_fn = Number;
+        } else if (!is_source_bigint && is_dest_bigint) {
+            // TypeError: Cannot convert [x] to a BigInt
+            map_fn = BigInt;
+        }
+
         // @ts-ignore
-        return new Tensor(type, DataTypeMap[type].from(this.data), this.dims);
+        return new Tensor(type, DataTypeMap[type].from(this.data, map_fn), this.dims);
     }
 }
 
@@ -33350,6 +33985,29 @@ async function topk(x, k) {
     });
 }
 
+
+const arrayToIndexTensor = (array) => new Tensor('int64', array, [array.length]);
+/**
+ * Slice a multidimensional float32 tensor.
+ * @param {Tensor} data: Tensor of data to extract slices from
+ * @param {number[]} starts: 1-D array of starting indices of corresponding axis in axes
+ * @param {number[]} ends: 1-D array of ending indices (exclusive) of corresponding axis in axes
+ * @param {number[]} axes: 1-D array of axes that starts and ends apply to
+ * @param {number[]} [steps]: 1-D array of slice step of corresponding axis in axes.
+ * @returns {Promise<Tensor>} Sliced data tensor.
+ */
+async function slice(data, starts, ends, axes, steps) {
+    const op = await _ops_registry_js__WEBPACK_IMPORTED_MODULE_2__.TensorOpRegistry.slice;
+    return await op({
+        x: data, 
+        s: arrayToIndexTensor(starts), 
+        e: arrayToIndexTensor(ends), 
+        a: arrayToIndexTensor(axes), 
+        t: arrayToIndexTensor(steps ?? new Array(axes.length).fill(1)),
+    });
+}
+
+
 /**
  * Perform mean pooling of the last hidden state followed by a normalization step.
  * @param {Tensor} last_hidden_state Tensor of shape [batchSize, seqLength, embedDim]
@@ -33796,6 +34454,20 @@ function zeros_like(tensor) {
     return zeros(tensor.dims);
 }
 
+/**
+ * Returns a tensor filled with random numbers from a uniform distribution on the interval [0, 1)
+ * @param {number[]} size A sequence of integers defining the shape of the output tensor.
+ * @returns {Tensor} The random tensor.
+ */
+function rand(size) {
+    const length = size.reduce((a, b) => a * b, 1);
+    return new Tensor(
+        "float32",
+        Float32Array.from({ length }, () => Math.random()),
+        size,
+    )
+}
+
 /**
  * Quantizes the embeddings tensor to binary or unsigned binary precision.
  * @param {Tensor} tensor The tensor to quantize.
@@ -34141,6 +34813,9 @@ __webpack_require__.r(__webpack_exports__);
 /* harmony export */   EsmModel: () => (/* reexport safe */ _models_js__WEBPACK_IMPORTED_MODULE_2__.EsmModel),
 /* harmony export */   EsmPreTrainedModel: () => (/* reexport safe */ _models_js__WEBPACK_IMPORTED_MODULE_2__.EsmPreTrainedModel),
 /* harmony export */   EsmTokenizer: () => (/* reexport safe */ _tokenizers_js__WEBPACK_IMPORTED_MODULE_3__.EsmTokenizer),
+/* harmony export */   ExaoneForCausalLM: () => (/* reexport safe */ _models_js__WEBPACK_IMPORTED_MODULE_2__.ExaoneForCausalLM),
+/* harmony export */   ExaoneModel: () => (/* reexport safe */ _models_js__WEBPACK_IMPORTED_MODULE_2__.ExaoneModel),
+/* harmony export */   ExaonePreTrainedModel: () => (/* reexport safe */ _models_js__WEBPACK_IMPORTED_MODULE_2__.ExaonePreTrainedModel),
 /* harmony export */   FFT: () => (/* reexport safe */ _utils_maths_js__WEBPACK_IMPORTED_MODULE_8__.FFT),
 /* harmony export */   FalconForCausalLM: () => (/* reexport safe */ _models_js__WEBPACK_IMPORTED_MODULE_2__.FalconForCausalLM),
 /* harmony export */   FalconModel: () => (/* reexport safe */ _models_js__WEBPACK_IMPORTED_MODULE_2__.FalconModel),
@@ -34319,7 +34994,17 @@ __webpack_require__.r(__webpack_exports__);
 /* harmony export */   MobileViTV2Model: () => (/* reexport safe */ _models_js__WEBPACK_IMPORTED_MODULE_2__.MobileViTV2Model),
 /* harmony export */   MobileViTV2PreTrainedModel: () => (/* reexport safe */ _models_js__WEBPACK_IMPORTED_MODULE_2__.MobileViTV2PreTrainedModel),
 /* harmony export */   ModelOutput: () => (/* reexport safe */ _models_js__WEBPACK_IMPORTED_MODULE_2__.ModelOutput),
+/* harmony export */   ModernBertForMaskedLM: () => (/* reexport safe */ _models_js__WEBPACK_IMPORTED_MODULE_2__.ModernBertForMaskedLM),
+/* harmony export */   ModernBertForSequenceClassification: () => (/* reexport safe */ _models_js__WEBPACK_IMPORTED_MODULE_2__.ModernBertForSequenceClassification),
+/* harmony export */   ModernBertForTokenClassification: () => (/* reexport safe */ _models_js__WEBPACK_IMPORTED_MODULE_2__.ModernBertForTokenClassification),
+/* harmony export */   ModernBertModel: () => (/* reexport safe */ _models_js__WEBPACK_IMPORTED_MODULE_2__.ModernBertModel),
+/* harmony export */   ModernBertPreTrainedModel: () => (/* reexport safe */ _models_js__WEBPACK_IMPORTED_MODULE_2__.ModernBertPreTrainedModel),
 /* harmony export */   Moondream1ForConditionalGeneration: () => (/* reexport safe */ _models_js__WEBPACK_IMPORTED_MODULE_2__.Moondream1ForConditionalGeneration),
+/* harmony export */   MoonshineFeatureExtractor: () => (/* reexport safe */ _models_feature_extractors_js__WEBPACK_IMPORTED_MODULE_10__.MoonshineFeatureExtractor),
+/* harmony export */   MoonshineForConditionalGeneration: () => (/* reexport safe */ _models_js__WEBPACK_IMPORTED_MODULE_2__.MoonshineForConditionalGeneration),
+/* harmony export */   MoonshineModel: () => (/* reexport safe */ _models_js__WEBPACK_IMPORTED_MODULE_2__.MoonshineModel),
+/* harmony export */   MoonshinePreTrainedModel: () => (/* reexport safe */ _models_js__WEBPACK_IMPORTED_MODULE_2__.MoonshinePreTrainedModel),
+/* harmony export */   MoonshineProcessor: () => (/* reexport safe */ _models_processors_js__WEBPACK_IMPORTED_MODULE_16__.MoonshineProcessor),
 /* harmony export */   MptForCausalLM: () => (/* reexport safe */ _models_js__WEBPACK_IMPORTED_MODULE_2__.MptForCausalLM),
 /* harmony export */   MptModel: () => (/* reexport safe */ _models_js__WEBPACK_IMPORTED_MODULE_2__.MptModel),
 /* harmony export */   MptPreTrainedModel: () => (/* reexport safe */ _models_js__WEBPACK_IMPORTED_MODULE_2__.MptPreTrainedModel),
@@ -34371,6 +35056,10 @@ __webpack_require__.r(__webpack_exports__);
 /* harmony export */   Phi3ForCausalLM: () => (/* reexport safe */ _models_js__WEBPACK_IMPORTED_MODULE_2__.Phi3ForCausalLM),
 /* harmony export */   Phi3Model: () => (/* reexport safe */ _models_js__WEBPACK_IMPORTED_MODULE_2__.Phi3Model),
 /* harmony export */   Phi3PreTrainedModel: () => (/* reexport safe */ _models_js__WEBPACK_IMPORTED_MODULE_2__.Phi3PreTrainedModel),
+/* harmony export */   Phi3VForCausalLM: () => (/* reexport safe */ _models_js__WEBPACK_IMPORTED_MODULE_2__.Phi3VForCausalLM),
+/* harmony export */   Phi3VImageProcessor: () => (/* reexport safe */ _models_image_processors_js__WEBPACK_IMPORTED_MODULE_13__.Phi3VImageProcessor),
+/* harmony export */   Phi3VPreTrainedModel: () => (/* reexport safe */ _models_js__WEBPACK_IMPORTED_MODULE_2__.Phi3VPreTrainedModel),
+/* harmony export */   Phi3VProcessor: () => (/* reexport safe */ _models_processors_js__WEBPACK_IMPORTED_MODULE_16__.Phi3VProcessor),
 /* harmony export */   PhiForCausalLM: () => (/* reexport safe */ _models_js__WEBPACK_IMPORTED_MODULE_2__.PhiForCausalLM),
 /* harmony export */   PhiModel: () => (/* reexport safe */ _models_js__WEBPACK_IMPORTED_MODULE_2__.PhiModel),
 /* harmony export */   PhiPreTrainedModel: () => (/* reexport safe */ _models_js__WEBPACK_IMPORTED_MODULE_2__.PhiPreTrainedModel),
@@ -34619,9 +35308,11 @@ __webpack_require__.r(__webpack_exports__);
 /* harmony export */   permute_data: () => (/* reexport safe */ _utils_maths_js__WEBPACK_IMPORTED_MODULE_8__.permute_data),
 /* harmony export */   pipeline: () => (/* reexport safe */ _pipelines_js__WEBPACK_IMPORTED_MODULE_1__.pipeline),
 /* harmony export */   quantize_embeddings: () => (/* reexport safe */ _utils_tensor_js__WEBPACK_IMPORTED_MODULE_7__.quantize_embeddings),
+/* harmony export */   rand: () => (/* reexport safe */ _utils_tensor_js__WEBPACK_IMPORTED_MODULE_7__.rand),
 /* harmony export */   read_audio: () => (/* reexport safe */ _utils_audio_js__WEBPACK_IMPORTED_MODULE_5__.read_audio),
 /* harmony export */   rfft: () => (/* reexport safe */ _utils_tensor_js__WEBPACK_IMPORTED_MODULE_7__.rfft),
 /* harmony export */   round: () => (/* reexport safe */ _utils_maths_js__WEBPACK_IMPORTED_MODULE_8__.round),
+/* harmony export */   slice: () => (/* reexport safe */ _utils_tensor_js__WEBPACK_IMPORTED_MODULE_7__.slice),
 /* harmony export */   softmax: () => (/* reexport safe */ _utils_maths_js__WEBPACK_IMPORTED_MODULE_8__.softmax),
 /* harmony export */   spectrogram: () => (/* reexport safe */ _utils_audio_js__WEBPACK_IMPORTED_MODULE_5__.spectrogram),
 /* harmony export */   stack: () => (/* reexport safe */ _utils_tensor_js__WEBPACK_IMPORTED_MODULE_7__.stack),