replace q8 with q5_K_M

app.py

@@ -10,7 +10,7 @@ MODEL_PATH = snapshot_download("nopperl/emissions-extraction-lora-merged-GGUF")
 
 def predict(input_method, document_file, document_url):
     document_path = document_file if input_method == "File" else document_url
-    emissions = extract_emissions(document_path, MODEL_PATH, model_name="ggml-model-q8_0.gguf")
+    emissions = extract_emissions(document_path, MODEL_PATH, model_name="ggml-model-Q5_K_M.gguf")
     return emissions.model_dump_json()
 
 with open("description.md", "r") as f:

article.md

@@ -1 +1 @@
-Technical overview: The system retrieves the relevant pages of the uploaded report using simple search. These pages are input into a finetuned [Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2) language model, which outputs a JSON object containing the emission information. The system achieves an emission extraction accuracy of 62% and a source citation accuracy of 67% on the [corporate-emission-reports](https://huggingface.co/datasets/nopperl/corporate-emission-reports) dataset. Note that the model is quantized due to resource limitations.
+Technical overview: The system retrieves the relevant pages of the uploaded report using simple search. These pages are input into a finetuned [Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2) language model, which outputs a JSON object containing the emission information. The system achieves an emission extraction accuracy of 65% and a source citation accuracy of 69% on the [corporate-emission-reports](https://huggingface.co/datasets/nopperl/corporate-emission-reports) dataset. Note that the model is quantized due to resource limitations.