Unknown model type: TF_LITE_AUDIO_ENCODER_HW

[qsj]

on MacBook pro m2
i download https://huggingface.co/google/gemma-3n-E2B-it-litert-lm/blob/main/gemma-3n-E2B-it-int4.litertlm directly
download https://github.com/google-ai-edge/LiteRT-LM/releases/latest/download/litert_lm_main.macos_arm64
store litert_lm_main.macos_arm64 and gemma-3n-E2B-it-int4.litertlm same directory

when i run
./litert_lm_main.macos_arm64 --backend=cpu --model_path=./gemma-3n-E2B-it-int4.litertlm --input_prompt="Write me a song"
WARNING: All log messages before absl::InitializeLog() is called are written to STDERR
I0000 00:00:1758683867.423778 15612525 litert_lm_main.cc:133] Model path: ./gemma-3n-E2B-it-int4.litertlm
I0000 00:00:1758683867.423951 15612525 litert_lm_main.cc:137] Choose backend: cpu
I0000 00:00:1758683867.424056 15612525 litert_lm_main.cc:144] executor_settings: backend: CPU
backend_config: number_of_threads: 4

max_tokens: 0
activation_data_type: Not set.
max_num_images: 0
cache_dir:
cache_file: Not set.
model_assets: model_path: ./gemma-3n-E2B-it-int4.litertlm
fake_weights_mode: FAKE_WEIGHTS_NONE

I0000 00:00:1758683867.424071 15612525 litert_lm_main.cc:155] Creating engine
I0000 00:00:1758683867.424163 15612525 litert_lm_loader.cc:121] LitertLmLoader::Initialize
I0000 00:00:1758683905.003587 15612525 litert_lm_loader.cc:128] mmap_status is ok
I0000 00:00:1758683905.003658 15612525 litert_lm_loader.cc:129] length: 3388604416
I0000 00:00:1758683905.004551 15612525 litert_lm_loader.cc:66] status: OK
I0000 00:00:1758683905.004567 15612525 litert_lm_loader.cc:67] major_version: 1
I0000 00:00:1758683905.004570 15612525 litert_lm_loader.cc:68] minor_version: 3
I0000 00:00:1758683905.004571 15612525 litert_lm_loader.cc:69] patch_version: 0
I0000 00:00:1758683905.004576 15612525 litert_lm_loader.cc:111] section_index: 0
I0000 00:00:1758683905.004577 15612525 litert_lm_loader.cc:112] section_data_type: LlmMetadataProto
I0000 00:00:1758683905.004578 15612525 litert_lm_loader.cc:114] section_begin_offset: 16384
I0000 00:00:1758683905.004580 15612525 litert_lm_loader.cc:115] section_end_offset: 16482
I0000 00:00:1758683905.004581 15612525 litert_lm_loader.cc:111] section_index: 1
I0000 00:00:1758683905.004583 15612525 litert_lm_loader.cc:112] section_data_type: SP_Tokenizer
I0000 00:00:1758683905.004584 15612525 litert_lm_loader.cc:114] section_begin_offset: 32768
I0000 00:00:1758683905.004585 15612525 litert_lm_loader.cc:115] section_end_offset: 4716087
I0000 00:00:1758683905.004587 15612525 litert_lm_loader.cc:97] model_type: TF_LITE_EMBEDDER
I0000 00:00:1758683905.004591 15612525 litert_lm_loader.cc:111] section_index: 2
I0000 00:00:1758683905.004592 15612525 litert_lm_loader.cc:112] section_data_type: TFLiteModel
I0000 00:00:1758683905.004593 15612525 litert_lm_loader.cc:114] section_begin_offset: 4718592
I0000 00:00:1758683905.004594 15612525 litert_lm_loader.cc:115] section_end_offset: 276303696
I0000 00:00:1758683905.004595 15612525 litert_lm_loader.cc:97] model_type: TF_LITE_PER_LAYER_EMBEDDER
I0000 00:00:1758683905.004597 15612525 litert_lm_loader.cc:111] section_index: 3
I0000 00:00:1758683905.004598 15612525 litert_lm_loader.cc:112] section_data_type: TFLiteModel
I0000 00:00:1758683905.004649 15612525 litert_lm_loader.cc:114] section_begin_offset: 276316160
I0000 00:00:1758683905.004703 15612525 litert_lm_loader.cc:115] section_end_offset: 1377335624
I0000 00:00:1758683905.004711 15612525 litert_lm_loader.cc:97] model_type: TF_LITE_AUDIO_ENCODER_HW
libc++abi: terminating due to uncaught exception of type absl::lts_20230802::BadStatusOrAccess: Bad StatusOr access: INVALID_ARGUMENT: Unknown model type: TF_LITE_AUDIO_ENCODER_HW

the same ERROR when i use gemma-3n-E4B-it-int4.litertlm

i also check the model sha256 its same with source in huggingface

[laianapinho]

I was using main branch, and I switched to stable v0.7.0 tag, maybe they are testing things in main.

[jonatlr]

Same behavior on RaspberryPi 5 (Raspberry OS). Using v0.7.0 did not help.