Merge remote-tracking branch 'upstream/master'

sync commit
intel · Aug 26, 2024 · 1a9f326 · 1a9f326
2 parents aad91eb + 02dc74d
commit 1a9f326
Show file tree

Hide file tree

Showing 12 changed files with 411 additions and 546 deletions.
diff --git a/README.md b/README.md
@@ -1,9 +1,11 @@
 # AI Innovation Bridge
 
-Welcome to our AI Innovation Bridge repo! This repo contains reference solutions, hackathon sample code, workshop content, and more!
+Welcome to our AI Innovation Bridge repo! This repo contains reference solutions, hackathon sample code, workshop content, and more!  
 
 ## Table of Contents
 - [hackathons](hackathons)
 - [solutions](solutions)
 - [workshops](workshops)
 - [utilities](utilities)
+
+Enjoy!
diff --git a/...023/pharmaceutical_manufacturing_business/medication_demand_forecast/src/requirements.txt b/...023/pharmaceutical_manufacturing_business/medication_demand_forecast/src/requirements.txt
@@ -1,7 +1,7 @@
 fastapi  
 uvicorn
-scikit-learn==1.0.1
-tensorflow==2.11.1
+scikit-learn==1.5.0
+tensorflow==2.12.1
 numpy
 pandas==1.4.1
 pydantic
diff --git a/hackathons/hack-for-good-2022/machine-learning-track/requirements.txt b/hackathons/hack-for-good-2022/machine-learning-track/requirements.txt
@@ -10,7 +10,7 @@ modin==0.15.2
 numpy==1.23.2
 pandas==1.4.3
 plotly==5.10.0
-scikit-learn==1.1.2
+scikit-learn==1.5.0
 scikit-learn-intelex==2021.6.3
 seaborn==0.11.2
-xgboost==1.6.1
+xgboost==1.6.1
diff --git a/utilities/model-card-tools/intel-extentions/transformers/requirements.txt b/utilities/model-card-tools/intel-extentions/transformers/requirements.txt
@@ -1,8 +1,8 @@
 accelerate==0.28.0
 einops==0.7.0
-intel-extension-for-pytorch==2.1.100
+intel-extension-for-pytorch==2.2.0
 intel-extension-for-transformers==1.2.2
 neural-speed==0.2
 neural-compressor==2.5
-torch==2.1.1
-transformers==4.34.1
+torch==2.2.0
+transformers==4.38.0
diff --git a/...ngface/1 - Introduction to Hugging Face - Understanding Model Loading and Inference.ipynb b/...ngface/1 - Introduction to Hugging Face - Understanding Model Loading and Inference.ipynb
@@ -49,10 +49,21 @@
    },
    "outputs": [],
    "source": [
-    "!source /opt/intel/oneapi/setvars.sh #comment out if not running on Intel Developer Cloud Jupyter\n",
-    "!pip install git+https://github.com/huggingface/transformers\n",
-    "!pip install torch==2.1.0\n",
-    "!pip install tiktoken==0.5.2"
+    "import sys\n",
+    "!{sys.executable} -m pip install transformers==4.44.0 --no-warn-script-location > /dev/null\n",
+    "!{sys.executable} -m pip install torch==2.1.0 --no-warn-script-location > /dev/null\n",
+    "!{sys.executable} -m pip install tiktoken==0.5.2 --no-warn-script-location > /dev/null"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "54f36f4e-9605-4e99-9154-f2e76488ec25",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# force restart kernel to pull latest environment\n",
+    "exit()"
    ]
   },
   {
@@ -61,7 +72,7 @@
    "metadata": {},
    "source": [
     "#### Model Loading\n",
-    "Here, we load the model and tokenizer from Hugging Face. We're using `AutoModelForCausalLM` and `AutoTokenizer` from the `transformers` library to load \"stabilityai/stablelm-2-1_6b\", a causal language model. The `trust_remote_code` flag is set to `True` for remote code execution, and `torch_dtype` is set to \"auto\" for automatic precision setting, optimizing performance on the underlying hardware..\n"
+    "Here, we load the model and tokenizer from Hugging Face. We're using `AutoModelForCausalLM` and `AutoTokenizer` from the `transformers` library to load \"microsoft/Phi-3.5-mini-instruct\", a causal language model. The `trust_remote_code` flag is set to `True` for remote code execution, and `torch_dtype` is set to \"auto\" for automatic precision setting, optimizing performance on the underlying hardware..\n"
    ]
   },
   {
@@ -74,10 +85,13 @@
    "outputs": [],
    "source": [
     "# Load model directly\n",
+    "\n",
+    "model = \"microsoft/Phi-3.5-mini-instruct\"\n",
+    "\n",
     "from transformers import AutoModelForCausalLM, AutoTokenizer\n",
-    "tokenizer = AutoTokenizer.from_pretrained(\"stabilityai/stablelm-2-1_6b\", trust_remote_code=True)\n",
+    "tokenizer = AutoTokenizer.from_pretrained(model, trust_remote_code=True)\n",
     "model = AutoModelForCausalLM.from_pretrained(\n",
-    "  \"stabilityai/stablelm-2-1_6b\",\n",
+    "  model,\n",
     "  trust_remote_code=True,\n",
     "  torch_dtype=\"auto\",\n",
     ")"
@@ -137,17 +151,7 @@
     "        torch_dtype=torch.bfloat16,\n",
     "        trust_remote_code=True,\n",
     "        device_map=\"auto\",\n",
-    "    )\n",
-    "\n",
-    "sequences = generator( \n",
-    "            \"I love learning to code...\",\n",
-    "            max_length=64,\n",
-    "            do_sample=True,\n",
-    "            top_k=3,\n",
-    "            temperature=0.70,\n",
-    "            top_p=0.95,\n",
-    "            num_return_sequences=1,\n",
-    "            eos_token_id=tokenizer.eos_token_id,)"
+    "    )"
    ]
   },
   {
@@ -159,6 +163,24 @@
     "In the final cell, we iterate through the sequences generated by our pipeline and print them out. This step is crucial for visualizing the outputs of our text generation pipeline, allowing us to evaluate the model's performance in generating coherent and contextually relevant text..\n"
    ]
   },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "d13188b2-4585-4524-8b4d-14450fef8bac",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "sequences = generator( \n",
+    "            \"I love learning to code...\",\n",
+    "            max_length=64,\n",
+    "            do_sample=True,\n",
+    "            top_k=3,\n",
+    "            temperature=0.70,\n",
+    "            top_p=0.95,\n",
+    "            num_return_sequences=1,\n",
+    "            eos_token_id=tokenizer.eos_token_id,)"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": null,
@@ -189,7 +211,7 @@
  ],
  "metadata": {
   "kernelspec": {
-   "display_name": "base",
+   "display_name": "Base",
    "language": "python",
    "name": "base"
   },
@@ -203,7 +225,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.9.16"
+   "version": "3.9.19"
   }
  },
  "nbformat": 4,

diff --git a/...2 - Leveraging Intel Optimizations with Hugging Face for Enhanced Model Performance.ipynb b/...2 - Leveraging Intel Optimizations with Hugging Face for Enhanced Model Performance.ipynb
@@ -8,8 +8,6 @@
    "source": [
     "# Leveraging Intel Optimizations with Hugging Face for Enhanced Model Performance\n",
     "\n",
-    "<img src=\"https://www.developer-tech.com/wp-content/uploads/sites/3/2023/08/intel-pytorch-foundation-ai-development-artificial-intelligence-coding-programming-machine-learning.jpg\" alt=\"Alt Text\" style=\"width: 400px;\"/>\n",
-    "\n",
     "Welcome to this developer-centric workshop where we explore the synergies between Hugging Face and Intel Extension for PyTorch (IPEX). This notebook serves as an introduction to utilizing IPEX for fine-tuning a pre-trained model, specifically focusing on the `distilbert-base-uncased` model for multi-class emotion classification in text.\n",
     "\n",
     "## Why This is Important\n",
@@ -44,12 +42,23 @@
    },
    "outputs": [],
    "source": [
-    "!source /opt/intel/oneapi/setvars.sh #comment out if not running on Intel Developer Cloud Jupyter\n",
-    "!pip install transformers==4.35.2\n",
-    "!pip install torch==2.1.0\n",
-    "!pip install intel_extension_for_pytorch==2.1.0\n",
-    "!pip install datasets==2.16.1\n",
-    "!pip install accelerate==0.26.0"
+    "import sys\n",
+    "!{sys.executable} -m pip install transformers==4.44.0 --no-warn-script-location > /dev/null\n",
+    "!{sys.executable} -m pip install torch==2.2.0 --no-warn-script-location > /dev/null\n",
+    "!{sys.executable} -m pip install intel_extension_for_pytorch==2.2.0 --no-warn-script-location > /dev/null\n",
+    "!{sys.executable} -m pip install datasets==2.16.1 --no-warn-script-location > /dev/null\n",
+    "!{sys.executable} -m pip install accelerate==0.32.0 --no-warn-script-location > /dev/null"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "0cb853ac-e1a6-41dc-9b68-da66b010e578",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# force restart kernel to pull latest environment\n",
+    "exit()"
    ]
   },
   {
@@ -177,14 +186,14 @@
     "# Define training arguments\n",
     "training_args = TrainingArguments(\n",
     "    output_dir=\"./results\",\n",
-    "    learning_rate=2e-5,\n",
+    "    learning_rate=2e-3,\n",
     "    per_device_train_batch_size=16,\n",
     "    per_device_eval_batch_size=16,\n",
     "    num_train_epochs=1,\n",
     "    weight_decay=0.01,\n",
     "    evaluation_strategy=\"epoch\",\n",
-    "    bf16=True, \n",
-    "    use_ipex=True,\n",
+    "    bf16=True, #False to train at FP32\n",
+    "    use_ipex=True, #False to train w/o IPEX\n",
     "    no_cuda=True,\n",
     ")\n",
     "\n",
@@ -288,7 +297,7 @@
  ],
  "metadata": {
   "kernelspec": {
-   "display_name": "base",
+   "display_name": "Base",
    "language": "python",
    "name": "base"
   },
@@ -302,7 +311,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.9.16"
+   "version": "3.9.19"
   }
  },
  "nbformat": 4,