update neuralchat notebook on cpu (#1496)

intel · Apr 19, 2024 · 83bb65a · 83bb65a
1 parent 37f9db2
commit 83bb65a
Show file tree

Hide file tree

Showing 14 changed files with 28 additions and 85 deletions.
diff --git a/intel_extension_for_transformers/neural_chat/docs/notebooks/amp_optimization_on_spr.ipynb b/intel_extension_for_transformers/neural_chat/docs/notebooks/amp_optimization_on_spr.ipynb
@@ -53,7 +53,7 @@
    "outputs": [],
    "source": [
     "%cd ./intel-extension-for-transformers/intel_extension_for_transformers/neural_chat/\n",
-    "!pip install -r requirements.txt\n",
+    "!pip install -r requirements_cpu.txt\n",
     "%cd ../../../"
    ]
   },

diff --git a/intel_extension_for_transformers/neural_chat/docs/notebooks/build_chatbot_on_icx.ipynb b/intel_extension_for_transformers/neural_chat/docs/notebooks/build_chatbot_on_icx.ipynb
@@ -53,29 +53,10 @@
    "outputs": [],
    "source": [
     "%cd ./intel-extension-for-transformers/intel_extension_for_transformers/neural_chat/\n",
-    "!pip install -r requirements.txt\n",
+    "!pip install -r requirements_cpu.txt\n",
     "%cd ../../../"
    ]
   },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "!pip uninstall torch -y\n",
-    "!pip install torch"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "!conda list"
-   ]
-  },
   {
    "cell_type": "markdown",
    "metadata": {},

diff --git a/intel_extension_for_transformers/neural_chat/docs/notebooks/build_chatbot_on_spr.ipynb b/intel_extension_for_transformers/neural_chat/docs/notebooks/build_chatbot_on_spr.ipynb
@@ -55,29 +55,10 @@
    "outputs": [],
    "source": [
     "%cd ./intel-extension-for-transformers/intel_extension_for_transformers/neural_chat/\n",
-    "!pip install -r requirements.txt\n",
+    "!pip install -r requirements_cpu.txt\n",
     "%cd ../../../"
    ]
   },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "!pip uninstall torch -y\n",
-    "!pip install torch"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "!conda list"
-   ]
-  },
   {
    "cell_type": "markdown",
    "metadata": {},

diff --git a/...for_transformers/neural_chat/docs/notebooks/customize_chatbot_with_finetuned_models.ipynb b/...for_transformers/neural_chat/docs/notebooks/customize_chatbot_with_finetuned_models.ipynb
@@ -109,7 +109,7 @@
     ")\n",
     "from intel_extension_for_transformers.neural_chat.chatbot import finetune_model\n",
     "model_args = ModelArguments(model_name_or_path=\"meta-llama/Llama-2-7b-chat-hf\")\n",
-    "data_args = DataArguments(train_file=\"alpaca_data.json\")\n",
+    "data_args = DataArguments(train_file=\"alpaca_data.json\", validation_split_percentage=1)\n",
     "training_args = TrainingArguments(\n",
     "    output_dir='./finetuned_model_path',\n",
     "    do_train=True,\n",

diff --git a/...xtension_for_transformers/neural_chat/docs/notebooks/customize_chatbot_with_plugins.ipynb b/...xtension_for_transformers/neural_chat/docs/notebooks/customize_chatbot_with_plugins.ipynb
@@ -58,16 +58,6 @@
     "%cd ../../../"
    ]
   },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "!pip uninstall torch -y\n",
-    "!pip install torch"
-   ]
-  },
   {
    "cell_type": "markdown",
    "metadata": {},
@@ -98,7 +88,9 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "With RAG Plugin"
+    "With RAG Plugin\n",
+    "\n",
+    "Note: Please prepare Xeon_8480_Processor.docx in your working directory."
    ]
   },
   {

diff --git a/intel_extension_for_transformers/neural_chat/docs/notebooks/deploy_chatbot_on_icx.ipynb b/intel_extension_for_transformers/neural_chat/docs/notebooks/deploy_chatbot_on_icx.ipynb
@@ -45,20 +45,10 @@
    "source": [
     "!git clone https://github.com/intel/intel-extension-for-transformers.git\n",
     "%cd ./intel-extension-for-transformers/intel_extension_for_transformers/neural_chat/\n",
-    "!pip install -r requirements.txt\n",
+    "!pip install -r requirements_cpu.txt\n",
     "%cd ../../../"
    ]
   },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "!pip uninstall torch -y\n",
-    "!pip install torch"
-   ]
-  },
   {
    "cell_type": "markdown",
    "metadata": {},

diff --git a/...on_for_transformers/neural_chat/docs/notebooks/finetune_neuralchat_v2_on_Nvidia_GPU.ipynb b/...on_for_transformers/neural_chat/docs/notebooks/finetune_neuralchat_v2_on_Nvidia_GPU.ipynb
@@ -21,9 +21,7 @@
     "!git clone https://github.com/intel/intel-extension-for-transformers.git\n",
     "%cd ./intel-extension-for-transformers/intel_extension_for_transformers/neural_chat/\n",
     "!pip install -r requirements.txt\n",
-    "%cd ../../../\n",
-    "!pip uninstall torch -y\n",
-    "!pip install torch"
+    "%cd ../../../"
    ]
   },
   {

diff --git a/intel_extension_for_transformers/neural_chat/docs/notebooks/finetuning_on_nv_a100.ipynb b/intel_extension_for_transformers/neural_chat/docs/notebooks/finetuning_on_nv_a100.ipynb
@@ -56,16 +56,6 @@
     "%cd ../../../"
    ]
   },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "!pip uninstall torch -y\n",
-    "!pip install torch"
-   ]
-  },
   {
    "cell_type": "markdown",
    "metadata": {},
@@ -112,7 +102,7 @@
     ")\n",
     "from intel_extension_for_transformers.neural_chat.chatbot import finetune_model\n",
     "model_args = ModelArguments(model_name_or_path=\"meta-llama/Llama-2-7b-chat-hf\")\n",
-    "data_args = DataArguments(train_file=\"alpaca_data.json\")\n",
+    "data_args = DataArguments(train_file=\"alpaca_data.json\", validation_split_percentage=1)\n",
     "training_args = TrainingArguments(\n",
     "    output_dir='./tmp',\n",
     "    do_train=True,\n",

diff --git a/...or_transformers/neural_chat/docs/notebooks/itrex_llm_graph_int4_optimization_on_spr.ipynb b/...or_transformers/neural_chat/docs/notebooks/itrex_llm_graph_int4_optimization_on_spr.ipynb
@@ -97,6 +97,15 @@
     "## INT4 Optimization"
    ]
   },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "!pip install gguf"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": null,

diff --git a/...on_for_transformers/neural_chat/docs/notebooks/setup_talking_chatbot_service_on_spr.ipynb b/...on_for_transformers/neural_chat/docs/notebooks/setup_talking_chatbot_service_on_spr.ipynb
@@ -30,7 +30,7 @@
     "!pip install intel-extension-for-transformers\n",
     "!git clone https://github.com/intel/intel-extension-for-transformers.git\n",
     "%cd ./intel-extension-for-transformers/intel_extension_for_transformers/neural_chat/\n",
-    "!pip install -r requirements.txt"
+    "!pip install -r requirements_cpu.txt"
    ]
   },
   {

diff --git a/..._for_transformers/neural_chat/docs/notebooks/setup_text_chatbot_with_caching_on_spr.ipynb b/..._for_transformers/neural_chat/docs/notebooks/setup_text_chatbot_with_caching_on_spr.ipynb
@@ -37,7 +37,7 @@
     "!pip install intel-extension-for-transformers\n",
     "!git clone https://github.com/intel/intel-extension-for-transformers.git\n",
     "%cd ./intel-extension-for-transformers/intel_extension_for_transformers/neural_chat/\n",
-    "!pip install -r requirements.txt\n",
+    "!pip install -r requirements_cpu.txt\n",
     "!sudo apt install numactl\n",
     "!conda install astunparse ninja pyyaml mkl mkl-include setuptools cmake cffi typing_extensions future six requests dataclasses -y\n",
     "!conda install jemalloc gperftools -c conda-forge -y\n",

diff --git a/...extension_for_transformers/neural_chat/docs/notebooks/single_node_finetuning_on_spr.ipynb b/...extension_for_transformers/neural_chat/docs/notebooks/single_node_finetuning_on_spr.ipynb
@@ -110,7 +110,7 @@
     ")\n",
     "from intel_extension_for_transformers.neural_chat.chatbot import finetune_model\n",
     "model_args = ModelArguments(model_name_or_path=\"meta-llama/Llama-2-7b-chat-hf\")\n",
-    "data_args = DataArguments(train_file=\"alpaca_data.json\")\n",
+    "data_args = DataArguments(train_file=\"alpaca_data.json\", validation_split_percentage=1)\n",
     "training_args = TrainingArguments(\n",
     "    output_dir='./tmp',\n",
     "    do_train=True,\n",

diff --git a/intel_extension_for_transformers/neural_chat/docs/notebooks/transformers_extension_api.ipynb b/intel_extension_for_transformers/neural_chat/docs/notebooks/transformers_extension_api.ipynb
@@ -28,7 +28,8 @@
    "outputs": [],
    "source": [
     "!pip install intel-extension-for-transformers\n",
-    "!pip install neural-speed"
+    "!pip install neural-speed\n",
+    "!pip install gguf"
    ]
   },
   {
@@ -82,13 +83,12 @@
     "\n",
     "# Download Hugging Face GPTQ model to local path\n",
     "model_name = \"PATH_TO_MODEL\"  # local path to model\n",
-    "woq_config = GPTQConfig(bits=4)\n",
     "prompt = \"Once upon a time, a little girl\"\n",
     "\n",
     "tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)\n",
     "inputs = tokenizer(prompt, return_tensors=\"pt\").input_ids\n",
     "streamer = TextStreamer(tokenizer)\n",
-    "model = AutoModelForCausalLM.from_pretrained(model_name, quantization_config=woq_config, trust_remote_code=True)\n",
+    "model = AutoModelForCausalLM.from_pretrained(model_name, trust_remote_code=True)\n",
     "outputs = model.generate(inputs, streamer=streamer, max_new_tokens=300)"
    ]
   },
@@ -117,7 +117,8 @@
     "sq_config = SmoothQuantConfig(\n",
     "                            tokenizer=tokenizer,  # either two of one, tokenizer or calib_func\n",
     "                            calib_iters=5,\n",
-    "                            recipes=recipes\n",
+    "                            recipes=recipes,\n",
+    "                            excluded_precisions = [\"bf16\"]\n",
     "                        )\n",
     "q_model = AutoModelForCausalLM.from_pretrained(\"Intel/neural-chat-7b-v3-1\",\n",
     "                                            quantization_config=sq_config,\n",

diff --git a/intel_extension_for_transformers/neural_chat/pipeline/plugins/audio/requirements.txt b/intel_extension_for_transformers/neural_chat/pipeline/plugins/audio/requirements.txt
@@ -5,6 +5,7 @@ inflect
 jaconv
 jieba
 langid
+librosa
 num2words
 openjtalk; sys_platform != 'linux'
 pydub