Update app.py

app.py

@@ -1,48 +1,88 @@
+# ADD DISCLAIMERS
+# AND LOGGING
+
+
 import gradio as gr
-from huggingface_hub import InferenceClient
 import os
-os.system("pip install transformers torch")
-from transformers import AutoTokenizer, TextIteratorStreamer, GenerationConfig, AutoModelForCausalLM
-import torch
+os.system("apt install libopenblas-dev")
+os.system("make clean && LLAMA_OPENBLAS=1 make")
+os.system('CMAKE_ARGS="-DLLAMA_BLAS=ON -DLLAMA_BLAS_VENDOR=OpenBLAS" pip install llama-cpp-python transformers')
+
+from llama_cpp import Llama
+from transformers import AutoTokenizer
+model_id = "Elijahbodden/eliGPTv1.1"
 
-model_id = "Elijahbodden/eliGPTv1.2"
+# MODEL
+model = Llama.from_pretrained(
+    repo_id=model_id",
+    filename="eliGPTv1.1-unsloth.Q5_K_M.gguf",
+    verbose=True,
+    n_threads = 2,
+    n_threads_batch = 2,
+    n_ctx=8192,
+  )
+
+
+# TOKENIZER AND TEMPLATE
 tokenizer = AutoTokenizer.from_pretrained(model_id)
-sober_model = AutoModelForCausalLM.from_pretrained(model_id)
 
-"""
-For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
-"""
-client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
+sys_prompt = """SUMMARY - ELIJAH:
+Age: 16
+Interests: space flight, cybernetics, consciousness, philosophy, psychonautism, biotech, AI
+Likes: thinking and learning, building stuff, interesting conversations, red hot chili peppers and techno, humanism
+Traits: incredibly intelligent, funny, interesting, caffeine fiend, very ambitious, militant atheist, self-taught/homeschooled, casual
+Aspirations: creating transhumanist utopia, understanding the universe more, becoming smarter and better\n"""
+
+your_name = "elijah"
+custom_template = \
+    f"{{% if {'sys_prompt' in locals()} %}}"\
+      "{{ '<|im_start|>system\n" + sys_prompt + "<|im_end|>\n' }}"\
+    "{% endif %}"\
+    "{% for message in messages %}"\
+        "{% if message['role'] == 'user' %}"\
+            "{{'<|im_start|>user\n' + message['content'] + '<|im_end|>\n'}}"\
+        "{% elif message['role'] == 'assistant' %}"\
+            "{{'<|im_start|>" + your_name + "\n' + message['content'] + eos_token + '<|im_end|>\n' }}"\
+        "{% else %}"\
+            "{{ '<|im_start|>system\n' + message['content'] + '<|im_end|>\n' }}"\
+        "{% endif %}"\
+    "{% endfor %}"\
+    "{{ '<|im_start|>" + your_name + "\n' }}"\
+
+tokenizer.chat_template = custom_template
 
 
 def respond(
     message,
     history: list[tuple[str, str]],
-    system_message,
     max_tokens,
     temperature,
-    top_p,
+    mirostat_tau,
+    mirostat_eta,
 ):
-    messages = [{"role": "system", "content": system_message}]
-
     for val in history:
         if val[0]:
-            messages.append({"role": "user", "content": val[0]})
+            messages.append({"from": "human", "content": val[0]})
         if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
+            messages.append({"from": "gpt", "content": val[1]})
 
-    messages.append({"role": "user", "content": message})
+    messages.append({"from": "human", "content": message})
 
     response = ""
 
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
+    tokenizer.apply_chat_template(messages, tokenize=False)
+    
+    for message in model.create_completion(
+        convo,
+        temperature=0.75,
         stream=True,
-        temperature=temperature,
-        top_p=top_p,
+        stop=["<|im_end|>"],
+        mirostat_mode=1,
+        mirostat_tau=mirostat_tau,
+        mirostat_eta=mirostat_eta,
+        max_tokens=128
     ):
-        token = message.choices[0].delta.content
+        token = message["choices"][0]["text"]
 
         response += token
         yield response
@@ -57,11 +97,18 @@ demo = gr.ChatInterface(
         gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
         gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
         gr.Slider(
-            minimum=0.1,
+            minimum=0.0,
+            maximum=10.0,
+            value=3.0,
+            step=0.5,
+            label="Mirostat tau",
+        ),
+        gr.Slider(
+            minimum=0.0,
             maximum=1.0,
-            value=0.95,
-            step=0.05,
-            label="Top-p (nucleus sampling)",
+            value=0.1,
+            step=0.01,
+            label="Mirostat eta",
         ),
     ],
 )