AttributeError: 'OmniSpeech2SLlamaForCausalLM' object has no attribute '_get_logits_warper'. Did you mean: '_get_logits_processor'?

[satheeshkola-532]

when i tried to infer the llama-omni s2s language model, getting the following error on giving the audio input, also same thing happend when tried using code.

AttributeError: 'OmniSpeech2SLlamaForCausalLM' object has no attribute '_get_logits_warper'. Did you mean: '_get_logits_processor'?

[Jintao-Huang]

Please try lowering the transformers version, e.g. 4.44.

[satheeshkola-532]

Please try lowering the transformers version, e.g. 4.44.

thank you, that error resolved but i am facing another issue which is that it's giving blank response as show in the image(even though i tried with the given examples), please let me know if there's any way to solve it and also i would like to get the python script to inference using swift, thank you!

[satheeshkola-532]

Please try lowering the transformers version, e.g. 4.44.

thank you, that error resolved but i am facing another issue which is that it's giving blank response as show in the image(even though i tried with the given examples), please let me know if there's any way to solve it and also i would like to get the python script to inference using swift, thank you!

@Jintao-Huang Any update on this issue?

[Jintao-Huang]

let me check. The holiday has just ended.

[Jintao-Huang]

fixed

[satheeshkola-532]

@Jintao-Huang > fixed

thank you @Jintao-Huang , it is generating text responses only on giving audio input but not both the text response and audio output ? Actually llama omni model generates both text and audio responses simultaneously right? Is there any way to achieve this?