[VLM] Support multimodal inputs for Florence-2 models (#13320)

This commit is contained in:
Isotr0py
2025-02-27 02:06:41 -08:00
committed by GitHub
parent 788f284b53
commit edf309ebbe
13 changed files with 1075 additions and 114 deletions
+1 -1
View File
@@ -105,7 +105,6 @@ _TEXT_GENERATION_MODELS = {
# [Encoder-decoder]
"BartModel": ("bart", "BartForConditionalGeneration"),
"BartForConditionalGeneration": ("bart", "BartForConditionalGeneration"),
"Florence2ForConditionalGeneration": ("florence2", "Florence2ForConditionalGeneration"), # noqa: E501
}
_EMBEDDING_MODELS = {
@@ -182,6 +181,7 @@ _MULTIMODAL_MODELS = {
"Qwen2AudioForConditionalGeneration": ("qwen2_audio", "Qwen2AudioForConditionalGeneration"), # noqa: E501
"UltravoxModel": ("ultravox", "UltravoxModel"),
# [Encoder-decoder]
"Florence2ForConditionalGeneration": ("florence2", "Florence2ForConditionalGeneration"), # noqa: E501
"MllamaForConditionalGeneration": ("mllama", "MllamaForConditionalGeneration"), # noqa: E501
"WhisperForConditionalGeneration": ("whisper", "WhisperForConditionalGeneration"), # noqa: E501
}