Elgato_dark/chronos-forecasting
1
0
Fork 0
mirror of https://github.com/amazon-science/chronos-forecasting synced 2026-05-24 01:58:27 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Fix links

Browse source
This commit is contained in:
Oleksandr Shchur 2025-12-29 17:15:57 +00:00
parent 649d1bc620
commit fe3256feb1
1 changed files with 8 additions and 7 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

15
notebooks/deploy-chronos-to-amazon-sagemaker.ipynb
View file

@ -21,22 +21,22 @@
"### Deployment Options\n",
"This notebook covers three deployment modes on SageMaker:\n",
"\n",
"**[Real-time Inference](https://docs.aws.amazon.com/sagemaker/latest/dg/realtime-endpoints.html)** [(Section 1)](#Section-1:-Real-time-Inference)\n",
"**[Real-time Inference](https://docs.aws.amazon.com/sagemaker/latest/dg/realtime-endpoints.html)** (Section 1)\n",
"- ✅ Highest throughput, consistently low latency, supports both GPU and CPU instances\n",
"- ✅ Simple setup via JumpStart\n",
"- ❌ By default, you pay for the time the endpoint is running (can be configured to [scale to zero](https://docs.aws.amazon.com/sagemaker/latest/dg/endpoint-auto-scaling-zero-instances.html))\n",
"\n",
"**[Serverless Inference](https://docs.aws.amazon.com/sagemaker/latest/dg/serverless-endpoints.html)** [(Section 2)](#Section-2:-Serverless-Inference)\n",
"**[Serverless Inference](https://docs.aws.amazon.com/sagemaker/latest/dg/serverless-endpoints.html)** (Section 2)\n",
"- ✅ Pay only for active inference time, no infrastructure management\n",
"- ✅ Cost-efficient for intermittent or unpredictable traffic\n",
"- ❌ Cold start latency on first request after idle, CPU only, lowest throughput of all options\n",
"- ❌ [More complex setup](#Setup-for-Serverless-and-Batch-Transform) (requires repackaging model artifacts)\n",
"- ❌ More complex setup (requires repackaging model artifacts)\n",
"\n",
"**[Batch Transform](https://docs.aws.amazon.com/sagemaker/latest/dg/batch-transform.html)** [(Section 3)](#Section-3:-Batch-Transform)\n",
"**[Batch Transform](https://docs.aws.amazon.com/sagemaker/latest/dg/batch-transform.html)** (Section 3)\n",
"- ✅ Pay only for active compute time, no persistent infrastructure\n",
"- ✅ Cost-efficient for large-scale batch prediction jobs\n",
"- ❌ Initialization takes severa minutes for each job (not for real-time use), CPU only, requires data in S3\n",
"- ❌ [More complex setup](#Setup-for-Serverless-and-Batch-Transform) (requires repackaging model artifacts)\n",
"- ❌ More complex setup (requires repackaging model artifacts)\n",
"\n",
"**Reference benchmark** on a dataset with 1M rows (2000 time series with 500 observations each) and prediction length of 28:\n",
"| Mode | Instance | Inference time (s) |\n",
@ -48,7 +48,7 @@
"\n",
"We recommend starting with **Real-time Inference** as it offers the simplest setup and highest throughput. Consider Serverless or Batch Transform when you need to optimize costs and don't require GPU acceleration.\n",
"\n",
"For a complete specification of all supported request parameters, see the [Endpoint API Reference](#Endpoint-API-Reference) at the end of this notebook."
"For a complete specification of all supported request parameters, see the **Endpoint API Reference** at the end of this notebook."
]
},
{
@ -1149,7 +1149,6 @@
"\n",
"[Batch Transform](https://docs.aws.amazon.com/sagemaker/latest/dg/batch-transform.html) processes large datasets offline. SageMaker spins up compute, processes all data, and shuts down automatically.\n",
"\n",
"The model uses the same API as described in the [Endpoint API Reference](#Endpoint-API-Reference), so you need to prepare your data in the expected JSON format.\n",
"\n",
"**When to use:**\n",
"- Large-scale batch forecasting (thousands of time series)\n",
@ -1168,6 +1167,8 @@
"source": [
"### Prepare Input Data\n",
"\n",
"The model uses the same API as described in the Endpoint API Reference at the end of the notebook, so you need to prepare your data in the expected JSON format.\n",
"\n",
"Batch Transform reads input from S3. Each line in the input file is a JSON payload that can contain multiple time series. For large datasets, use `items_per_record` to control how many time series are included per line (and thus per request)."
]
},