Batchalign 2 - Unable to run Whisper Engine / Forced Alignment

[RichardGhiuzan]

I have been using the library version of Batchalign 2 to produce transcripts, and while I've been able to do so, I've been getting a recurrent error whenever trying to run a forced alignment on an existing transcript using the following line of coding:

chat = ba.CHATFile(path = "inputfile.cha") doc = chat.doc nlp = ba.BatchalignPipeline.new("fa", lang="eng", num_speakers=2) doc = nlp(chat) final = ba.CHATFile(doc=doc) final.write("final.cha")

I get a very lengthy error output, which I can share if needed, but this seems to be the main error: RuntimeError Traceback (most recent call last) Cell In[38], line 4 2 chat = ba.CHATFile(path = "inputfile.cha") 3 doc = chat.doc ----> 4 nlp = ba.BatchalignPipeline.new("fa", lang="eng", num_speakers=2) 5 doc = nlp(chat) 6 final = ba.CHATFile(doc=doc)

[...]

2572 raise ValueError( 2573 "You cannot cast a GPTQ model in a new dtype. Make sure to load the model using from_pretrained using the desired" 2574 " dtype by passing the correct torch_dtype argument."

Would anyone have encountered a similar error and know how to solve it? Thank you in advance for the time and availability. Kind Regards, Richard

[Jemoka]

Apologies about this! Looking at your script, could you try:

chat = ba.CHATFile(path = "inputfile.cha")
doc = chat.doc
nlp = ba.BatchalignPipeline.new("fa", lang="eng", num_speakers=2)
# this is wrong: ~~doc = nlp(chat)~~
doc = nlp(doc)
final = ba.CHATFile(doc=doc)
final.write("final.cha")

However, something is really odd on that none of this should involve GPTQ in any way.