Upload folder using huggingface_hub

.gradio/certificate.pem

@@ -0,0 +1,31 @@
+-----BEGIN CERTIFICATE-----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=
+-----END CERTIFICATE-----

README.md

@@ -1,12 +1,13 @@
 ---
-title: Llama Rjochtwurd
-emoji: 💻
-colorFrom: purple
-colorTo: green
+title: Rjochtwurd
+emoji: 🐢
+colorFrom: red
+colorTo: indigo
 sdk: gradio
 sdk_version: 5.6.0
 app_file: app.py
 pinned: false
+short_description: Frisian Text-to-Speech Error Correction
 ---
 
 Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

app.py

@@ -0,0 +1,73 @@
+from transformers import pipeline
+from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig
+import torch
+import gradio as gr
+
+from peft import PeftModel
+
+lora_wheights = "slugroom/llama_rjochtwurd"
+base_model_id = "meta-llama/Llama-3.2-3B"
+
+tokenizer = AutoTokenizer.from_pretrained(base_model_id)
+
+# to save the model and tokenizer to huggingface hub
+print("saving model and tokenizer")
+
+model = AutoModelForCausalLM.from_pretrained(base_model_id)
+model = PeftModel.from_pretrained(model, lora_wheights, torch_dtype=torch.float16, quantization_config=BitsAndBytesConfig(load_in_8bit=True)).to("cpu")
+
+
+# model_id = "slugroom/llama_rjochtwurd"
+# tokenizer = AutoTokenizer.from_pretrained(model_id)
+# model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype=torch.float16).to("cpu")
+
+asr = pipeline("automatic-speech-recognition", "wietsedv/wav2vec2-large-xlsr-53-frisian", device="cpu")
+
+def speech_to_text(speech):
+    text = asr(speech)["text"]  
+    return text
+
+def correct_errors(text):
+    in_txt = f"The following is a Frisian audio transcription, some parts of the transcription may be incorrect. Correct the transcription by making it grammatically and phonetically accurate.\n ### Transcription: {text} \n ### Corrected:"
+    model_inputs = tokenizer([in_txt], return_tensors="pt")
+    
+    generated_ids = model.generate(**model_inputs, max_new_tokens=100, eos_token_id=tokenizer.eos_token_id, pad_token_id=tokenizer.pad_token_id, do_sample=True, temperature=0.5, top_k=50)
+
+    # remove prompt from generated text
+    generated_ids = generated_ids[:, model_inputs["input_ids"].shape[-1]:]
+
+    corrected = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
+
+    print(corrected)
+    if "###" in corrected:
+        corrected = corrected.split("###")[0]
+
+
+    return corrected
+
+
+demo = gr.Blocks()
+
+with demo:
+    gr.HTML("<h1>Frision Speech-to-Text Error Correction</h1>")
+    gr.HTML("<p>Upload an audio file in Frisian and press 'Transcribe Audio' to transcribe the audio. The transcription will appear in the text box below. You can also type Frisian text in the text box to be corrected and press 'Correct Errors' to correct the text.</p>")
+    gr.HTML("<p>Note that correcting the errors may take up to a minute.</p>")
+    
+    audio_file = gr.Audio(type="filepath", label="Upload Audio File")
+    gr.Examples([["./samples/1.mp3"], ["./samples/2.mp3"], ["./samples/3.mp3"], ["./samples/4.mp3"]], audio_file, label="Example Audio Files")
+    text = gr.Textbox(placeholder="Transcription will appear here... Additionally type Frisian here to be corrected", label="Transcription")
+    gr.Examples([["ik wit net wêrt ûnder skoraaf"],["hiy it is mei jammer"], ["ast myn frou de earenijen hinne e litte wol is it hare saak"], ["dêr wiene hiel wat sasjalisten yn fryslân"]], text, label="Examples of Frisian Transcriptions")
+    audio_file.upload(fn=speech_to_text, inputs=audio_file, outputs=text)
+
+    corrected_text = gr.Textbox(interactive=False, label="Corrected Transcription")
+
+    with gr.Row():
+        b1 = gr.Button("Transcribe Audio")
+        b1.click(speech_to_text, inputs=audio_file, outputs=text)
+        
+        b2 = gr.Button("Correct Errors")
+        b2.click(correct_errors, inputs=text, outputs=corrected_text)
+
+demo.launch(server_port=5000)
+
+

requirements.txt

@@ -0,0 +1,10 @@
+torch==2.5.1
+torchaudio==2.5.1
+torchvision==0.20.1
+transformers==4.46.2
+numpy==1.26.4
+gradio==5.6.0
+bitsandbytes==0.42.0
+PySoundFile
+llama-stack
+peft==0.13.2