use updated quantized model

app.py

@@ -3,13 +3,13 @@ from llama_cpp import Llama
 
 MAX_TOKENS = 64
 
-llm = Llama(model_path="ggml-model-f16-q4_0.bin", n_ctx=2048)
+llm = Llama(model_path="ggml-model.bin", n_ctx=2048)
 
 def generate_text_instruct(input_text):
     response = ""
     txt2tag_prompt = f"You are a tool that helps tag danbooru images when given a textual image description. Provide me with danbooru tags that accurately fit the following description. {input_text}"
     for output in llm(f" A chat between a curious user and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the user's questions. USER: {txt2tag_prompt} ASSISTANT:", 
-                      echo=False, stream=True, max_tokens=64, stop=["</s>", "\n", "User:", "<unk>"]):
+                      echo=False, stream=True, max_tokens=96, stop=["</s>", "\n", "User:", "<unk>"]):
         answer = output['choices'][0]['text']
         response += answer
         yield response

ggml-model-f16-q4_0.bin → ggml-model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a214d083617ae4cb2971d73235043d3505407017cd0d98639ee762618e09c177
-size 7323305088
+oid sha256:f92ee3dd8e5d103896669b0c38c4e63d6f7f87a099ad98b325683476d0cc9c14
+size 9763701888