"
+ ]
+ },
+ "metadata": {},
+ "output_type": "display_data"
+ },
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "Keyboard interruption in main thread... closing server.\n"
+ ]
+ },
+ {
+ "data": {
+ "text/html": [
+ "╭─────────────────────────────── Traceback (most recent call last) ────────────────────────────────╮\n",
+ "│ /usr/local/lib/python3.10/dist-packages/gradio/blocks.py:2058 in block_thread │\n",
+ "│ │\n",
+ "│ 2055 │ │ \"\"\"Block main thread until interrupted by user.\"\"\" │\n",
+ "│ 2056 │ │ try: │\n",
+ "│ 2057 │ │ │ while True: │\n",
+ "│ ❱ 2058 │ │ │ │ time.sleep(0.1) │\n",
+ "│ 2059 │ │ except (KeyboardInterrupt, OSError): │\n",
+ "│ 2060 │ │ │ print(\"Keyboard interruption in main thread... closing server.\") │\n",
+ "│ 2061 │ │ │ self.server.close() │\n",
+ "╰──────────────────────────────────────────────────────────────────────────────────────────────────╯\n",
+ "KeyboardInterrupt\n",
+ "\n",
+ "During handling of the above exception, another exception occurred:\n",
+ "\n",
+ "╭─────────────────────────────── Traceback (most recent call last) ────────────────────────────────╮\n",
+ "│ in <cell line: 236>:236 │\n",
+ "│ │\n",
+ "│ /usr/local/lib/python3.10/dist-packages/gradio/blocks.py:1970 in launch │\n",
+ "│ │\n",
+ "│ 1967 │ │ │\n",
+ "│ 1968 │ │ # Block main thread if debug==True │\n",
+ "│ 1969 │ │ if debug or int(os.getenv(\"GRADIO_DEBUG\", 0)) == 1: │\n",
+ "│ ❱ 1970 │ │ │ self.block_thread() │\n",
+ "│ 1971 │ │ # Block main thread if running in a script to stop script from exiting │\n",
+ "│ 1972 │ │ is_in_interactive_mode = bool(getattr(sys, \"ps1\", sys.flags.interactive)) │\n",
+ "│ 1973 │\n",
+ "│ │\n",
+ "│ /usr/local/lib/python3.10/dist-packages/gradio/blocks.py:2061 in block_thread │\n",
+ "│ │\n",
+ "│ 2058 │ │ │ │ time.sleep(0.1) │\n",
+ "│ 2059 │ │ except (KeyboardInterrupt, OSError): │\n",
+ "│ 2060 │ │ │ print(\"Keyboard interruption in main thread... closing server.\") │\n",
+ "│ ❱ 2061 │ │ │ self.server.close() │\n",
+ "│ 2062 │ │ │ for tunnel in CURRENT_TUNNELS: │\n",
+ "│ 2063 │ │ │ │ tunnel.kill() │\n",
+ "│ 2064 │\n",
+ "│ │\n",
+ "│ /usr/local/lib/python3.10/dist-packages/gradio/networking.py:43 in close │\n",
+ "│ │\n",
+ "│ 40 │ │\n",
+ "│ 41 │ def close(self): │\n",
+ "│ 42 │ │ self.should_exit = True │\n",
+ "│ ❱ 43 │ │ self.thread.join() │\n",
+ "│ 44 │\n",
+ "│ 45 │\n",
+ "│ 46 def get_first_available_port(initial: int, final: int) -> int: │\n",
+ "│ │\n",
+ "│ /usr/lib/python3.10/threading.py:1096 in join │\n",
+ "│ │\n",
+ "│ 1093 │ │ │ raise RuntimeError(\"cannot join current thread\") │\n",
+ "│ 1094 │ │ │\n",
+ "│ 1095 │ │ if timeout is None: │\n",
+ "│ ❱ 1096 │ │ │ self._wait_for_tstate_lock() │\n",
+ "│ 1097 │ │ else: │\n",
+ "│ 1098 │ │ │ # the behavior of a negative timeout isn't documented, but │\n",
+ "│ 1099 │ │ │ # historically .join(timeout=x) for x<0 has acted as if timeout=0 │\n",
+ "│ │\n",
+ "│ /usr/lib/python3.10/threading.py:1116 in _wait_for_tstate_lock │\n",
+ "│ │\n",
+ "│ 1113 │ │ │ return │\n",
+ "│ 1114 │ │ │\n",
+ "│ 1115 │ │ try: │\n",
+ "│ ❱ 1116 │ │ │ if lock.acquire(block, timeout): │\n",
+ "│ 1117 │ │ │ │ lock.release() │\n",
+ "│ 1118 │ │ │ │ self._stop() │\n",
+ "│ 1119 │ │ except: │\n",
+ "╰──────────────────────────────────────────────────────────────────────────────────────────────────╯\n",
+ "KeyboardInterrupt\n",
+ "
\n"
+ ],
+ "text/plain": [
+ "\u001b[31m╭─\u001b[0m\u001b[31m──────────────────────────────\u001b[0m\u001b[31m \u001b[0m\u001b[1;31mTraceback \u001b[0m\u001b[1;2;31m(most recent call last)\u001b[0m\u001b[31m \u001b[0m\u001b[31m───────────────────────────────\u001b[0m\u001b[31m─╮\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2;33m/usr/local/lib/python3.10/dist-packages/gradio/\u001b[0m\u001b[1;33mblocks.py\u001b[0m:\u001b[94m2058\u001b[0m in \u001b[92mblock_thread\u001b[0m \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m2055 \u001b[0m\u001b[2;90m│ │ \u001b[0m\u001b[33m\"\"\"Block main thread until interrupted by user.\"\"\"\u001b[0m \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m2056 \u001b[0m\u001b[2m│ │ \u001b[0m\u001b[94mtry\u001b[0m: \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m2057 \u001b[0m\u001b[2m│ │ │ \u001b[0m\u001b[94mwhile\u001b[0m \u001b[94mTrue\u001b[0m: \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[31m❱ \u001b[0m2058 \u001b[2m│ │ │ │ \u001b[0mtime.sleep(\u001b[94m0.1\u001b[0m) \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m2059 \u001b[0m\u001b[2m│ │ \u001b[0m\u001b[94mexcept\u001b[0m (\u001b[96mKeyboardInterrupt\u001b[0m, \u001b[96mOSError\u001b[0m): \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m2060 \u001b[0m\u001b[2m│ │ │ \u001b[0m\u001b[96mprint\u001b[0m(\u001b[33m\"\u001b[0m\u001b[33mKeyboard interruption in main thread... closing server.\u001b[0m\u001b[33m\"\u001b[0m) \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m2061 \u001b[0m\u001b[2m│ │ │ \u001b[0m\u001b[96mself\u001b[0m.server.close() \u001b[31m│\u001b[0m\n",
+ "\u001b[31m╰──────────────────────────────────────────────────────────────────────────────────────────────────╯\u001b[0m\n",
+ "\u001b[1;91mKeyboardInterrupt\u001b[0m\n",
+ "\n",
+ "\u001b[3mDuring handling of the above exception, another exception occurred:\u001b[0m\n",
+ "\n",
+ "\u001b[31m╭─\u001b[0m\u001b[31m──────────────────────────────\u001b[0m\u001b[31m \u001b[0m\u001b[1;31mTraceback \u001b[0m\u001b[1;2;31m(most recent call last)\u001b[0m\u001b[31m \u001b[0m\u001b[31m───────────────────────────────\u001b[0m\u001b[31m─╮\u001b[0m\n",
+ "\u001b[31m│\u001b[0m in \u001b[92m\u001b[0m:\u001b[94m236\u001b[0m \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2;33m/usr/local/lib/python3.10/dist-packages/gradio/\u001b[0m\u001b[1;33mblocks.py\u001b[0m:\u001b[94m1970\u001b[0m in \u001b[92mlaunch\u001b[0m \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m1967 \u001b[0m\u001b[2m│ │ \u001b[0m \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m1968 \u001b[0m\u001b[2m│ │ \u001b[0m\u001b[2m# Block main thread if debug==True\u001b[0m \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m1969 \u001b[0m\u001b[2m│ │ \u001b[0m\u001b[94mif\u001b[0m debug \u001b[95mor\u001b[0m \u001b[96mint\u001b[0m(os.getenv(\u001b[33m\"\u001b[0m\u001b[33mGRADIO_DEBUG\u001b[0m\u001b[33m\"\u001b[0m, \u001b[94m0\u001b[0m)) == \u001b[94m1\u001b[0m: \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[31m❱ \u001b[0m1970 \u001b[2m│ │ │ \u001b[0m\u001b[96mself\u001b[0m.block_thread() \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m1971 \u001b[0m\u001b[2m│ │ \u001b[0m\u001b[2m# Block main thread if running in a script to stop script from exiting\u001b[0m \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m1972 \u001b[0m\u001b[2m│ │ \u001b[0mis_in_interactive_mode = \u001b[96mbool\u001b[0m(\u001b[96mgetattr\u001b[0m(sys, \u001b[33m\"\u001b[0m\u001b[33mps1\u001b[0m\u001b[33m\"\u001b[0m, sys.flags.interactive)) \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m1973 \u001b[0m \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2;33m/usr/local/lib/python3.10/dist-packages/gradio/\u001b[0m\u001b[1;33mblocks.py\u001b[0m:\u001b[94m2061\u001b[0m in \u001b[92mblock_thread\u001b[0m \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m2058 \u001b[0m\u001b[2m│ │ │ │ \u001b[0mtime.sleep(\u001b[94m0.1\u001b[0m) \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m2059 \u001b[0m\u001b[2m│ │ \u001b[0m\u001b[94mexcept\u001b[0m (\u001b[96mKeyboardInterrupt\u001b[0m, \u001b[96mOSError\u001b[0m): \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m2060 \u001b[0m\u001b[2m│ │ │ \u001b[0m\u001b[96mprint\u001b[0m(\u001b[33m\"\u001b[0m\u001b[33mKeyboard interruption in main thread... closing server.\u001b[0m\u001b[33m\"\u001b[0m) \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[31m❱ \u001b[0m2061 \u001b[2m│ │ │ \u001b[0m\u001b[96mself\u001b[0m.server.close() \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m2062 \u001b[0m\u001b[2m│ │ │ \u001b[0m\u001b[94mfor\u001b[0m tunnel \u001b[95min\u001b[0m CURRENT_TUNNELS: \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m2063 \u001b[0m\u001b[2m│ │ │ │ \u001b[0mtunnel.kill() \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m2064 \u001b[0m \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2;33m/usr/local/lib/python3.10/dist-packages/gradio/\u001b[0m\u001b[1;33mnetworking.py\u001b[0m:\u001b[94m43\u001b[0m in \u001b[92mclose\u001b[0m \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m 40 \u001b[0m\u001b[2m│ \u001b[0m \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m 41 \u001b[0m\u001b[2m│ \u001b[0m\u001b[94mdef\u001b[0m \u001b[92mclose\u001b[0m(\u001b[96mself\u001b[0m): \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m 42 \u001b[0m\u001b[2m│ │ \u001b[0m\u001b[96mself\u001b[0m.should_exit = \u001b[94mTrue\u001b[0m \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[31m❱ \u001b[0m 43 \u001b[2m│ │ \u001b[0m\u001b[96mself\u001b[0m.thread.join() \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m 44 \u001b[0m \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m 45 \u001b[0m \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m 46 \u001b[0m\u001b[94mdef\u001b[0m \u001b[92mget_first_available_port\u001b[0m(initial: \u001b[96mint\u001b[0m, final: \u001b[96mint\u001b[0m) -> \u001b[96mint\u001b[0m: \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2;33m/usr/lib/python3.10/\u001b[0m\u001b[1;33mthreading.py\u001b[0m:\u001b[94m1096\u001b[0m in \u001b[92mjoin\u001b[0m \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m1093 \u001b[0m\u001b[2m│ │ │ \u001b[0m\u001b[94mraise\u001b[0m \u001b[96mRuntimeError\u001b[0m(\u001b[33m\"\u001b[0m\u001b[33mcannot join current thread\u001b[0m\u001b[33m\"\u001b[0m) \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m1094 \u001b[0m\u001b[2m│ │ \u001b[0m \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m1095 \u001b[0m\u001b[2m│ │ \u001b[0m\u001b[94mif\u001b[0m timeout \u001b[95mis\u001b[0m \u001b[94mNone\u001b[0m: \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[31m❱ \u001b[0m1096 \u001b[2m│ │ │ \u001b[0m\u001b[96mself\u001b[0m._wait_for_tstate_lock() \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m1097 \u001b[0m\u001b[2m│ │ \u001b[0m\u001b[94melse\u001b[0m: \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m1098 \u001b[0m\u001b[2m│ │ │ \u001b[0m\u001b[2m# the behavior of a negative timeout isn't documented, but\u001b[0m \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m1099 \u001b[0m\u001b[2m│ │ │ \u001b[0m\u001b[2m# historically .join(timeout=x) for x<0 has acted as if timeout=0\u001b[0m \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2;33m/usr/lib/python3.10/\u001b[0m\u001b[1;33mthreading.py\u001b[0m:\u001b[94m1116\u001b[0m in \u001b[92m_wait_for_tstate_lock\u001b[0m \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m1113 \u001b[0m\u001b[2m│ │ │ \u001b[0m\u001b[94mreturn\u001b[0m \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m1114 \u001b[0m\u001b[2m│ │ \u001b[0m \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m1115 \u001b[0m\u001b[2m│ │ \u001b[0m\u001b[94mtry\u001b[0m: \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[31m❱ \u001b[0m1116 \u001b[2m│ │ │ \u001b[0m\u001b[94mif\u001b[0m lock.acquire(block, timeout): \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m1117 \u001b[0m\u001b[2m│ │ │ │ \u001b[0mlock.release() \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m1118 \u001b[0m\u001b[2m│ │ │ │ \u001b[0m\u001b[96mself\u001b[0m._stop() \u001b[31m│\u001b[0m\n",
+ "\u001b[31m│\u001b[0m \u001b[2m1119 \u001b[0m\u001b[2m│ │ \u001b[0m\u001b[94mexcept\u001b[0m: \u001b[31m│\u001b[0m\n",
+ "\u001b[31m╰──────────────────────────────────────────────────────────────────────────────────────────────────╯\u001b[0m\n",
+ "\u001b[1;91mKeyboardInterrupt\u001b[0m\n"
+ ]
+ },
+ "metadata": {},
+ "output_type": "display_data"
+ }
+ ],
+ "source": [
+ "import gradio as gr\n",
+ "from langchain.vectorstores import FAISS\n",
+ "from langchain.embeddings import HuggingFaceEmbeddings\n",
+ "from transformers import LlamaTokenizer, LlamaForCausalLM, GenerationConfig, pipeline\n",
+ "from transformers import BitsAndBytesConfig\n",
+ "import textwrap\n",
+ "import torch\n",
+ "\n",
+ "quant_config = BitsAndBytesConfig(load_in_4bit=True,\n",
+ " bnb_4bit_compute_dtype=torch.bfloat16,\n",
+ " bnb_4bit_use_double_quant=True,\n",
+ " )\n",
+ "\n",
+ "prompt = 'BEGINNING OF CONVERSATION: USER: \\\n",
+ "I will provide you with two abstracts, I intend to use the author of the second to review the first. Tell me in 15 words why or why not the second author is a good fit to review the first paper.\\n\\\n",
+ "Abstract To Be Reviewed: '\n",
+ "\n",
+ "tokenizer = LlamaTokenizer.from_pretrained(\"samwit/koala-7b\")\n",
+ "\n",
+ "base_model = LlamaForCausalLM.from_pretrained(\n",
+ " \"samwit/koala-7b\",\n",
+ " load_in_8bit=False,\n",
+ " load_in_4bit=True,\n",
+ " device_map='auto',\n",
+ " offload_folder='offloaded',\n",
+ " quantization_config=quant_config\n",
+ ")\n",
+ "\n",
+ "pipe = pipeline(\n",
+ " \"text-generation\",\n",
+ " model=base_model,\n",
+ " tokenizer=tokenizer,\n",
+ " max_length=1024,\n",
+ " temperature=0.7,\n",
+ " top_p=0.95,\n",
+ " repetition_penalty=1.15,\n",
+ " # device=-1\n",
+ ")\n",
+ "\n",
+ "\n",
+ "def wrap_text_preserve_newlines(text, width=110):\n",
+ " # Split the input text into lines based on newline characters\n",
+ " lines = text.split('\\n')\n",
+ " # Wrap each line individually\n",
+ " wrapped_lines = [textwrap.fill(line, width=width) for line in lines]\n",
+ " # Join the wrapped lines back together using newline characters\n",
+ " wrapped_text = '\\n'.join(wrapped_lines)\n",
+ " return wrapped_text\n",
+ "\n",
+ "\n",
+ "def create_miread_embed(sents, bundle):\n",
+ " tokenizer = bundle[0]\n",
+ " model = bundle[1]\n",
+ " model.cpu()\n",
+ " tokens = tokenizer(sents,\n",
+ " max_length=512,\n",
+ " padding=True,\n",
+ " truncation=True,\n",
+ " return_tensors=\"pt\"\n",
+ " )\n",
+ " device = torch.device('cpu')\n",
+ " tokens = tokens.to(device)\n",
+ " with torch.no_grad():\n",
+ " out = model.bert(**tokens)\n",
+ " feature = out.last_hidden_state[:, 0, :]\n",
+ " return feature.cpu()\n",
+ "\n",
+ "\n",
+ "def get_matches(query, k):\n",
+ " matches = vecdb.similarity_search_with_score(query, k=k)\n",
+ " return matches\n",
+ "\n",
+ "\n",
+ "def inference(query,k=30,mode=''):\n",
+ " matches = get_matches(query,k)\n",
+ " j_bucket = {}\n",
+ " n_table = []\n",
+ " a_table = []\n",
+ " r_table = []\n",
+ " scores = [round(match[1].item(),3) for match in matches]\n",
+ " min_score = min(scores)\n",
+ " max_score = max(scores)\n",
+ " normaliser = lambda x: round(1 - (x-min_score)/max_score,3)\n",
+ " for i,match in enumerate(matches):\n",
+ " doc = match[0]\n",
+ " score = normaliser(round(match[1].item(),3))\n",
+ " title = doc.metadata['title']\n",
+ " author = doc.metadata['authors'][0]\n",
+ " date = doc.metadata.get('date','None')\n",
+ " link = doc.metadata.get('link','None')\n",
+ " submitter = doc.metadata.get('submitter','None')\n",
+ " journal = doc.metadata.get('journal','None')\n",
+ " abstract = doc.metadata.get('abstract','')\n",
+ "\n",
+ " # For journals\n",
+ " if journal not in j_bucket:\n",
+ " j_bucket[journal] = score\n",
+ " else:\n",
+ " j_bucket[journal] += score\n",
+ "\n",
+ " # For authors\n",
+ " record = [i+1,\n",
+ " score,\n",
+ " author,\n",
+ " title,\n",
+ " link,\n",
+ " date]\n",
+ " n_table.append(record)\n",
+ "\n",
+ " # For abstracts\n",
+ " record = [i+1,\n",
+ " title,\n",
+ " author,\n",
+ " submitter,\n",
+ " journal,\n",
+ " date,\n",
+ " link,\n",
+ " score\n",
+ " ]\n",
+ " a_table.append(record)\n",
+ "\n",
+ " # For reviewer\n",
+ " r_record = [i+1,\n",
+ " score,\n",
+ " author,\n",
+ " abstract,\n",
+ " title,\n",
+ " link,\n",
+ " date]\n",
+ " r_table.append(r_record)\n",
+ "\n",
+ "\n",
+ "\n",
+ " if (mode):\n",
+ " dataset = [prompt + query + '\\n Candidate Abstract: ' + row[3] + '\\n GPT:' for row in r_table[:5]]\n",
+ " outputs = pipe(dataset)\n",
+ " outputs = [output[0]['generated_text'].split('GPT:')[1] for output in outputs]\n",
+ " r_table = [[r[0],r[1],r[2],outputs[i],r[4],r[5],r[6]] for i,r in enumerate(r_table[:5])]\n",
+ " # print(f\"{i}/5 done\")\n",
+ " else:\n",
+ " outputs = ['']*5\n",
+ " r_table = [[r[0],r[1],r[2],outputs[i],r[4],r[5],r[6]] for i,r in enumerate(r_table[:5])]\n",
+ "\n",
+ " j_table = sorted([[journal,score] for journal,score in j_bucket.items()],key= lambda x : x[1],reverse=True)\n",
+ " j_table = [[i+1,item[0],item[1]] for i,item in enumerate(j_table)]\n",
+ " j_output= gr.Dataframe.update(value=j_table,visible=True)\n",
+ " n_output= gr.Dataframe.update(value=n_table,visible=True)\n",
+ " a_output = gr.Dataframe.update(value=a_table,visible=True)\n",
+ " if mode:\n",
+ " r_output = gr.Dataframe.update(value=r_table,visible=True)\n",
+ " return r_output\n",
+ "\n",
+ " return [a_output,j_output,n_output]\n",
+ "\n",
+ "def k_inference(query,k):\n",
+ " return inference(query,k,'koala')\n",
+ "\n",
+ "model_name = \"biodatlab/MIReAD-Neuro\"\n",
+ "model_kwargs = {'device': 'cpu'}\n",
+ "encode_kwargs = {'normalize_embeddings': False}\n",
+ "faiss_embedder = HuggingFaceEmbeddings(\n",
+ " model_name=model_name,\n",
+ " model_kwargs=model_kwargs,\n",
+ " encode_kwargs=encode_kwargs\n",
+ ")\n",
+ "\n",
+ "vecdb = FAISS.load_local(\"nbdt_data\", faiss_embedder)\n",
+ "\n",
+ "\n",
+ "with gr.Blocks(theme=gr.themes.Soft()) as demo:\n",
+ " gr.Markdown(\"# NBDT Recommendation Engine for Editors\")\n",
+ " gr.Markdown(\"NBDT Recommendation Engine for Editors is a tool for neuroscience authors/abstracts/journalsrecommendation built for NBDT journal editors. \\\n",
+ " It aims to help an editor to find similar reviewers, abstracts, and journals to a given submitted abstract.\\\n",
+ " To find a recommendation, paste a `title[SEP]abstract` or `abstract` in the text box below and click \\\"Find Matches\\\".\\\n",
+ " Then, you can hover to authors/abstracts/journals tab to find a suggested list.\\\n",
+ " The data in our current demo is selected from 2018 to 2022. We will update the data monthly for an up-to-date publications.\")\n",
+ "\n",
+ "\n",
+ " abst = gr.Textbox(label=\"Abstract\",lines=10)\n",
+ "\n",
+ " k = gr.Slider(1,100,step=1,value=50,label=\"Number of matches to consider\")\n",
+ "\n",
+ " action_btn = gr.Button(value=\"Find Matches\")\n",
+ "\n",
+ " with gr.Tab(\"Authors\"):\n",
+ " n_output = gr.Dataframe(\n",
+ " headers=['No.','Score','Name','Title','Link','Date'],\n",
+ " datatype=['number','number','str','str','str','str'],\n",
+ " col_count=(6, \"fixed\"),\n",
+ " wrap=True,\n",
+ " visible=False\n",
+ " )\n",
+ " with gr.Tab(\"Abstracts\"):\n",
+ " a_output = gr.Dataframe(\n",
+ " headers=['No.','Title','Author','Corresponding Author','Journal','Date','Link','Score'],\n",
+ " datatype=['number','str','str','str','str','str','str','number'],\n",
+ " col_count=(8,\"fixed\"),\n",
+ " wrap=True,\n",
+ " visible=False\n",
+ " )\n",
+ " with gr.Tab(\"Journals\"):\n",
+ " j_output = gr.Dataframe(\n",
+ " headers=['No.','Name','Score'],\n",
+ " datatype=['number','str','number'],\n",
+ " col_count=(3, \"fixed\"),\n",
+ " wrap=True,\n",
+ " visible=False\n",
+ " )\n",
+ "\n",
+ " llm_btn = gr.Button(value=\"Listen to Koala's advice\")\n",
+ " with gr.Tab(\"Reviewers (New)\"):\n",
+ " r_output = gr.Dataframe(\n",
+ " headers=['No.','Score','Name','Title','Reasoning','Link','Date'],\n",
+ " datatype=['number','number','str','str','str','str','str'],\n",
+ " col_count=(7,\"fixed\"),\n",
+ " wrap=True,\n",
+ " visible=True\n",
+ " )\n",
+ "\n",
+ " action_btn.click(fn=inference,\n",
+ " inputs=[\n",
+ " abst,\n",
+ " k,\n",
+ " # modes,\n",
+ " ],\n",
+ " outputs=[a_output,j_output,n_output],\n",
+ " api_name=\"neurojane\")\n",
+ " llm_btn.click(fn=k_inference,\n",
+ " inputs=[\n",
+ " abst,\n",
+ " k,\n",
+ " ],\n",
+ " outputs = [r_output],\n",
+ " api_name=\"koala\")\n",
+ "\n",
+ "demo.launch(debug=True,share=True)"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 30,
+ "metadata": {
+ "colab": {
+ "base_uri": "https://localhost:8080/"
+ },
+ "id": "UQhTTPJGBdwo",
+ "outputId": "93a00886-db64-4b6a-88b5-51f2d88c58e5"
+ },
+ "outputs": [
+ {
+ "data": {
+ "text/plain": [
+ "{'max_rows': None,\n",
+ " 'max_cols': None,\n",
+ " 'label': None,\n",
+ " 'show_label': None,\n",
+ " 'scale': None,\n",
+ " 'min_width': None,\n",
+ " 'interactive': None,\n",
+ " 'visible': True,\n",
+ " 'value': [[1,\n",
+ " 1.0,\n",
+ " 'philipp berens',\n",
+ " \" Both authors have expertise in neuroscience and machine learning, making them suitable candidates to review each other's work. The first author's research focuses on understanding the mechanisms underlying learning and plasticity, while the second author's work involves developing new algorithms for spike inference from calcium signals. These areas of expertise complement each other, as the first author's insights into the nature of plasticity can inform the development of more effective spike inference algorithms.\",\n",
+ " 'Supervised learning sets benchmark for robust spike detection from calcium imaging signals',\n",
+ " 'https://www.semanticscholar.org/paper/786f9831900ca34d8200291e5a72d2f14ad9b336',\n",
+ " 2015],\n",
+ " [2,\n",
+ " 0.974,\n",
+ " 'daniel butts',\n",
+ " \" Both authors have expertise in neuroscience and machine learning, making them suitable candidates to review each other's work. The first author's research focuses on understanding the mechanisms underlying learning and plasticity, while the second author's work uses information theory to analyze neural ensembles. While both authors may have different perspectives and approaches, they share a common goal of advancing our understanding of the brain and its functions.\",\n",
+ " 'Information Theoretic Measure of Stimulus Significance is Not Confounded by Stimulus Correlations or Non-Linearities',\n",
+ " 'https://www.semanticscholar.org/paper/f6767047e84872f4cb09f94c26b1d21e3ef9d55d',\n",
+ " 2002],\n",
+ " [3,\n",
+ " 0.952,\n",
+ " 'alex williams',\n",
+ " ' The candidate abstract provides a novel approach to understanding neural sequence data using a point process model that captures both the timing and occurrence of neural activity. This approach addresses some of the limitations of previous methods, including the need for spike times to be discretized and the use of a sub-optimal least-squares criterion. Additionally, the proposed method allows for the incorporation of learned time warping parameters, which have been shown to be effective in modeling variable sequence lengths. Overall, this approach has potential applications in understanding various aspects of neural processing, including working memory, motor production, and learning.',\n",
+ " 'Point process models for sequence detection in high-dimensional neural spike trains',\n",
+ " 'https://www.semanticscholar.org/paper/2942ed449bc35f06170342b61ebb6fc2dd6183c9',\n",
+ " 2020],\n",
+ " [4,\n",
+ " 0.931,\n",
+ " 'philipp berens',\n",
+ " \" Both authors have expertise in neuroscience and machine learning, making them suitable candidates to review each other's work. The first author's research focuses on understanding the mechanisms underlying learning and plasticity, while the second author's work involves developing new algorithms for spike rate inference in calcium imaging studies.\",\n",
+ " 'Supervised learning sets benchmark for robust spike rate inference from calcium imaging signals',\n",
+ " 'https://www.semanticscholar.org/paper/ff3c6755a17026d234978095dcc36a43b551ac89',\n",
+ " 2015],\n",
+ " [5,\n",
+ " 0.924,\n",
+ " 'konrad kording',\n",
+ " \" Both authors have expertise in the field of artificial intelligence and machine learning, specifically in the area of deep learning. They also have experience in theoretical neuroscience and cognitive psychology. However, the candidate's research focuses more specifically on the topic of perceptual processing and the use of adversarial algorithms as a potential mechanism for learning in the brain. On the other hand, the author of the original abstract reviews the existing literature on plasticity-related mechanisms and argues that gradients are a key concept for understanding the process of neuronal plasticity. While both authors have valuable insights into the topic, it may be beneficial to consult with experts in both fields to gain a comprehensive understanding of the subject matter.\",\n",
+ " 'Learning to infer in recurrent biological networks',\n",
+ " 'https://www.semanticscholar.org/paper/482085ff18eaefb262b8217797e99a2cca339938',\n",
+ " 2020]],\n",
+ " '__type__': 'update'}"
+ ]
+ },
+ "execution_count": 30,
+ "metadata": {},
+ "output_type": "execute_result"
+ }
+ ],
+ "source": [
+ "k_inference('The experimental study of learning and plasticity has always been driven by an implicit question: how can physiological changes be adaptive and improve performance? For example, in Hebbian plasticity only synapses from presynaptic neurons that were active are changed, avoiding useless changes. Similarly, in dopamine-gated learning synapse changes depend on reward or lack thereof and do not change when everything is predictable. Within machine learning we can make the question of which changes are adaptive concrete: performance improves when changes correlate with the gradient of an objective function quantifying performance. This result is general for any system that improves through small changes. As such, physiology has always implicitly been seeking mechanisms that allow the brain to approximate gradients. Coming from this perspective we review the existing literature on plasticity-related mechanisms, and we show how these mechanisms relate to gradient estimation. We argue that gradients are a unifying idea to explain the many facets of neuronal plasticity.',k=10)"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "metadata": {
+ "id": "Nxi4msI_aIl2"
+ },
+ "outputs": [],
+ "source": []
+ }
+ ],
+ "metadata": {
+ "accelerator": "GPU",
+ "colab": {
+ "provenance": []
+ },
+ "kernelspec": {
+ "display_name": "Python 3",
+ "name": "python3"
+ },
+ "language_info": {
+ "name": "python"
+ },
+ "widgets": {
+ "application/vnd.jupyter.widget-state+json": {
+ "05a91c8da3054845bff97967fb17372d": {
+ "model_module": "@jupyter-widgets/base",
+ "model_module_version": "1.2.0",
+ "model_name": "LayoutModel",
+ "state": {
+ "_model_module": "@jupyter-widgets/base",
+ "_model_module_version": "1.2.0",
+ "_model_name": "LayoutModel",
+ "_view_count": null,
+ "_view_module": "@jupyter-widgets/base",
+ "_view_module_version": "1.2.0",
+ "_view_name": "LayoutView",
+ "align_content": null,
+ "align_items": null,
+ "align_self": null,
+ "border": null,
+ "bottom": null,
+ "display": null,
+ "flex": null,
+ "flex_flow": null,
+ "grid_area": null,
+ "grid_auto_columns": null,
+ "grid_auto_flow": null,
+ "grid_auto_rows": null,
+ "grid_column": null,
+ "grid_gap": null,
+ "grid_row": null,
+ "grid_template_areas": null,
+ "grid_template_columns": null,
+ "grid_template_rows": null,
+ "height": null,
+ "justify_content": null,
+ "justify_items": null,
+ "left": null,
+ "margin": null,
+ "max_height": null,
+ "max_width": null,
+ "min_height": null,
+ "min_width": null,
+ "object_fit": null,
+ "object_position": null,
+ "order": null,
+ "overflow": null,
+ "overflow_x": null,
+ "overflow_y": null,
+ "padding": null,
+ "right": null,
+ "top": null,
+ "visibility": null,
+ "width": null
+ }
+ },
+ "435b61fe3135482587f1dcc65306e114": {
+ "model_module": "@jupyter-widgets/base",
+ "model_module_version": "1.2.0",
+ "model_name": "LayoutModel",
+ "state": {
+ "_model_module": "@jupyter-widgets/base",
+ "_model_module_version": "1.2.0",
+ "_model_name": "LayoutModel",
+ "_view_count": null,
+ "_view_module": "@jupyter-widgets/base",
+ "_view_module_version": "1.2.0",
+ "_view_name": "LayoutView",
+ "align_content": null,
+ "align_items": null,
+ "align_self": null,
+ "border": null,
+ "bottom": null,
+ "display": null,
+ "flex": null,
+ "flex_flow": null,
+ "grid_area": null,
+ "grid_auto_columns": null,
+ "grid_auto_flow": null,
+ "grid_auto_rows": null,
+ "grid_column": null,
+ "grid_gap": null,
+ "grid_row": null,
+ "grid_template_areas": null,
+ "grid_template_columns": null,
+ "grid_template_rows": null,
+ "height": null,
+ "justify_content": null,
+ "justify_items": null,
+ "left": null,
+ "margin": null,
+ "max_height": null,
+ "max_width": null,
+ "min_height": null,
+ "min_width": null,
+ "object_fit": null,
+ "object_position": null,
+ "order": null,
+ "overflow": null,
+ "overflow_x": null,
+ "overflow_y": null,
+ "padding": null,
+ "right": null,
+ "top": null,
+ "visibility": null,
+ "width": null
+ }
+ },
+ "56796deef330429d9a9fec4432ca54d4": {
+ "model_module": "@jupyter-widgets/controls",
+ "model_module_version": "1.5.0",
+ "model_name": "HTMLModel",
+ "state": {
+ "_dom_classes": [],
+ "_model_module": "@jupyter-widgets/controls",
+ "_model_module_version": "1.5.0",
+ "_model_name": "HTMLModel",
+ "_view_count": null,
+ "_view_module": "@jupyter-widgets/controls",
+ "_view_module_version": "1.5.0",
+ "_view_name": "HTMLView",
+ "description": "",
+ "description_tooltip": null,
+ "layout": "IPY_MODEL_05a91c8da3054845bff97967fb17372d",
+ "placeholder": "",
+ "style": "IPY_MODEL_64906f3cba5c433baae65ecaf06612af",
+ "value": " 14/14 [02:48<00:00, 11.14s/it]"
+ }
+ },
+ "64906f3cba5c433baae65ecaf06612af": {
+ "model_module": "@jupyter-widgets/controls",
+ "model_module_version": "1.5.0",
+ "model_name": "DescriptionStyleModel",
+ "state": {
+ "_model_module": "@jupyter-widgets/controls",
+ "_model_module_version": "1.5.0",
+ "_model_name": "DescriptionStyleModel",
+ "_view_count": null,
+ "_view_module": "@jupyter-widgets/base",
+ "_view_module_version": "1.2.0",
+ "_view_name": "StyleView",
+ "description_width": ""
+ }
+ },
+ "6885776cbe9449f8864676e1e83b4c19": {
+ "model_module": "@jupyter-widgets/controls",
+ "model_module_version": "1.5.0",
+ "model_name": "DescriptionStyleModel",
+ "state": {
+ "_model_module": "@jupyter-widgets/controls",
+ "_model_module_version": "1.5.0",
+ "_model_name": "DescriptionStyleModel",
+ "_view_count": null,
+ "_view_module": "@jupyter-widgets/base",
+ "_view_module_version": "1.2.0",
+ "_view_name": "StyleView",
+ "description_width": ""
+ }
+ },
+ "68db08a240724f9fb38f464db4c73b84": {
+ "model_module": "@jupyter-widgets/controls",
+ "model_module_version": "1.5.0",
+ "model_name": "HTMLModel",
+ "state": {
+ "_dom_classes": [],
+ "_model_module": "@jupyter-widgets/controls",
+ "_model_module_version": "1.5.0",
+ "_model_name": "HTMLModel",
+ "_view_count": null,
+ "_view_module": "@jupyter-widgets/controls",
+ "_view_module_version": "1.5.0",
+ "_view_name": "HTMLView",
+ "description": "",
+ "description_tooltip": null,
+ "layout": "IPY_MODEL_d83e0c6ff5db4a29a05fed5b72773be7",
+ "placeholder": "",
+ "style": "IPY_MODEL_6885776cbe9449f8864676e1e83b4c19",
+ "value": "Loading checkpoint shards: 100%"
+ }
+ },
+ "7530c375556442249c0f19ea435a629f": {
+ "model_module": "@jupyter-widgets/controls",
+ "model_module_version": "1.5.0",
+ "model_name": "ProgressStyleModel",
+ "state": {
+ "_model_module": "@jupyter-widgets/controls",
+ "_model_module_version": "1.5.0",
+ "_model_name": "ProgressStyleModel",
+ "_view_count": null,
+ "_view_module": "@jupyter-widgets/base",
+ "_view_module_version": "1.2.0",
+ "_view_name": "StyleView",
+ "bar_color": null,
+ "description_width": ""
+ }
+ },
+ "7da5f39db40f465caea5dd36ce881237": {
+ "model_module": "@jupyter-widgets/controls",
+ "model_module_version": "1.5.0",
+ "model_name": "FloatProgressModel",
+ "state": {
+ "_dom_classes": [],
+ "_model_module": "@jupyter-widgets/controls",
+ "_model_module_version": "1.5.0",
+ "_model_name": "FloatProgressModel",
+ "_view_count": null,
+ "_view_module": "@jupyter-widgets/controls",
+ "_view_module_version": "1.5.0",
+ "_view_name": "ProgressView",
+ "bar_style": "success",
+ "description": "",
+ "description_tooltip": null,
+ "layout": "IPY_MODEL_f32bc75b18d140aaba55de24947dbb3f",
+ "max": 14,
+ "min": 0,
+ "orientation": "horizontal",
+ "style": "IPY_MODEL_7530c375556442249c0f19ea435a629f",
+ "value": 14
+ }
+ },
+ "b427d9e2d2ee4f828ee896a03e9f38eb": {
+ "model_module": "@jupyter-widgets/controls",
+ "model_module_version": "1.5.0",
+ "model_name": "HBoxModel",
+ "state": {
+ "_dom_classes": [],
+ "_model_module": "@jupyter-widgets/controls",
+ "_model_module_version": "1.5.0",
+ "_model_name": "HBoxModel",
+ "_view_count": null,
+ "_view_module": "@jupyter-widgets/controls",
+ "_view_module_version": "1.5.0",
+ "_view_name": "HBoxView",
+ "box_style": "",
+ "children": [
+ "IPY_MODEL_68db08a240724f9fb38f464db4c73b84",
+ "IPY_MODEL_7da5f39db40f465caea5dd36ce881237",
+ "IPY_MODEL_56796deef330429d9a9fec4432ca54d4"
+ ],
+ "layout": "IPY_MODEL_435b61fe3135482587f1dcc65306e114"
+ }
+ },
+ "d83e0c6ff5db4a29a05fed5b72773be7": {
+ "model_module": "@jupyter-widgets/base",
+ "model_module_version": "1.2.0",
+ "model_name": "LayoutModel",
+ "state": {
+ "_model_module": "@jupyter-widgets/base",
+ "_model_module_version": "1.2.0",
+ "_model_name": "LayoutModel",
+ "_view_count": null,
+ "_view_module": "@jupyter-widgets/base",
+ "_view_module_version": "1.2.0",
+ "_view_name": "LayoutView",
+ "align_content": null,
+ "align_items": null,
+ "align_self": null,
+ "border": null,
+ "bottom": null,
+ "display": null,
+ "flex": null,
+ "flex_flow": null,
+ "grid_area": null,
+ "grid_auto_columns": null,
+ "grid_auto_flow": null,
+ "grid_auto_rows": null,
+ "grid_column": null,
+ "grid_gap": null,
+ "grid_row": null,
+ "grid_template_areas": null,
+ "grid_template_columns": null,
+ "grid_template_rows": null,
+ "height": null,
+ "justify_content": null,
+ "justify_items": null,
+ "left": null,
+ "margin": null,
+ "max_height": null,
+ "max_width": null,
+ "min_height": null,
+ "min_width": null,
+ "object_fit": null,
+ "object_position": null,
+ "order": null,
+ "overflow": null,
+ "overflow_x": null,
+ "overflow_y": null,
+ "padding": null,
+ "right": null,
+ "top": null,
+ "visibility": null,
+ "width": null
+ }
+ },
+ "f32bc75b18d140aaba55de24947dbb3f": {
+ "model_module": "@jupyter-widgets/base",
+ "model_module_version": "1.2.0",
+ "model_name": "LayoutModel",
+ "state": {
+ "_model_module": "@jupyter-widgets/base",
+ "_model_module_version": "1.2.0",
+ "_model_name": "LayoutModel",
+ "_view_count": null,
+ "_view_module": "@jupyter-widgets/base",
+ "_view_module_version": "1.2.0",
+ "_view_name": "LayoutView",
+ "align_content": null,
+ "align_items": null,
+ "align_self": null,
+ "border": null,
+ "bottom": null,
+ "display": null,
+ "flex": null,
+ "flex_flow": null,
+ "grid_area": null,
+ "grid_auto_columns": null,
+ "grid_auto_flow": null,
+ "grid_auto_rows": null,
+ "grid_column": null,
+ "grid_gap": null,
+ "grid_row": null,
+ "grid_template_areas": null,
+ "grid_template_columns": null,
+ "grid_template_rows": null,
+ "height": null,
+ "justify_content": null,
+ "justify_items": null,
+ "left": null,
+ "margin": null,
+ "max_height": null,
+ "max_width": null,
+ "min_height": null,
+ "min_width": null,
+ "object_fit": null,
+ "object_position": null,
+ "order": null,
+ "overflow": null,
+ "overflow_x": null,
+ "overflow_y": null,
+ "padding": null,
+ "right": null,
+ "top": null,
+ "visibility": null,
+ "width": null
+ }
+ }
+ }
+ }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 0
+}
|