Neuron SDK 2.13.0 Release Notes and new samples

Author: natemail-aws

README.md

@@ -14,7 +14,7 @@ Samples are organized by use case (training, inference) and deep learning framew
 
 | Usage | Description | Instance Type |
 | --- | --- | --- |
-| [Megatron-LM for Neuron](https://github.com/aws-neuron/aws-neuron-reference-for-megatron-lm) | A library that enables large-scale distributed training of language models such as GPT and is adapted from Megatron-LM. | Trn1, Trn1n |
+| [Nemo Megatron for Neuron](https://github.com/aws-neuron/neuronx-nemo-megatron) | A library that enables large-scale distributed training of language models such as Llama and is adapted from Nemo Megatron. | Trn1, Trn1n |
 | [AWS Neuron samples for ParallelCluster](https://github.com/aws-neuron/aws-neuron-parallelcluster-samples) | How to use AWS ParallelCluster to build HPC compute cluster that uses trn1 compute nodes to run your distributed ML training job.  | Trn1, Trn1n |
 | [AWS Neuron samples for EKS](https://github.com/aws-neuron/aws-neuron-eks-samples) | The samples in this repository demonstrate the types of patterns that can be used to deliver inference and distributed training on EKS using Inferentia and Trainium. | Trn1, Trn1n |
 | [AWS Neuron samples for SageMaker](https://github.com/aws-neuron/aws-neuron-sagemaker-samples) | SageMaker Samples using ml.trn1 instances for machine learning (ML) training workloads on the AWS ML accelerator chips Trainium. | Trn1, Trn1n |

releasenotes.md

@@ -1,5 +1,17 @@
 # Change Log
 
+## August, 28th 2023
+* Added sample script for LLaMA V2 13B model inference using transformers-neuronx
+* Added samples for training GPT-NEOX 20B and 6.9B models using neuronx-distributed 
+* Added sample scripts for CLIP and Stable Diffusion XL inference using torch-neuronx
+* Added sample scripts for vision and language Perceiver models inference using torch-neuronx
+* Added camembert training/finetuning example for Trn1 under hf_text_classification in torch-neuronx
+* Updated Fine-tuning Hugging Face BERT Japanese model sample in torch-neuronx
+* Updated OPT and GPT-J transformers-neuronx inference samples to install transformers-neuronx from whl instead of using github repo
+* Upgraded numpy package to 1.21.6 in GPT-2 and several training samples under hf_text_classification in torch-neuronx
+* Removed pinning of torch-neuron and tensorflow-neuron libraries and other minor changes in several of torch-neuron and tensorflow-neuron Inf1 inference samples.
+
+
 ## February, 23rd 2023
 * Added OPT-13B, OPT-30B, OPT-66B inference examples under transformers-neuronx
 * Added distilbert-base-uncased training/finetuning example for Trn1 under torch-neuronx

tensorflow-neuron/inference/unet/UnetTF2.ipynb

@@ -20,15 +20,12 @@
    ]
   },
   {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "d1081826",
+   "attachments": {},
+   "cell_type": "markdown",
+   "id": "b79b1ad3",
    "metadata": {},
-   "outputs": [],
    "source": [
-    "# Set Pip repository  to point to the Neuron repository\n",
-    "%pip config set global.extra-index-url https://pip.repos.neuron.amazonaws.com\n",
-    "# now restart the kernel"
+    "Verify that this Jupyter notebook is running the Python kernel environment that was set up according to the [Tensforlow Installation Guide](https://awsdocs-neuron.readthedocs-hosted.com/en/latest/general/setup/tensorflow-neuron.html#setup-tensorflow-neuron). You can select the kernel from the \"Kernel -> Change Kernel\" option on the top of this Jupyter notebook page."
    ]
   },
   {
@@ -41,7 +38,7 @@
    "outputs": [],
    "source": [
     "#Install Neuron Tensorflow\n",
-    "%pip install -U tensorflow-neuron==2.5.2.2.1.14.0 neuron-cc matplotlib\n",
+    "%pip install -U tensorflow-neuron neuron-cc matplotlib\n",
     "# use --force-reinstall if you're facing some issues while loading the modules\n",
     "# now restart the kernel again"
    ]
@@ -206,7 +203,7 @@
     "x = next(iter(x))[0]\n",
     "\n",
     "# https://awsdocs-neuron.readthedocs-hosted.com/en/latest/neuron-guide/neuron-cc/command-line-reference.html\n",
-    "os.environ[\"NEURON_CC_FLAGS\"] = \"--verbose=DEBUG --neuroncore-pipeline-cores=1 --workdir=logs/ --dynamic-batch-size\"\n",
+    "os.environ[\"NEURON_CC_FLAGS\"] = \"--neuroncore-pipeline-cores=1 --workdir=logs/ --dynamic-batch-size\"\n",
     "neuron_model = tfn.trace(unet_model, x)\n",
     "neuron_model.save('unet_circles_neuron')"
    ]