Sentiment Analysis Model

Runs inside your own AWS account: your text is processed on a SageMaker endpoint you control, inside your VPC, and is never sent to a third-party API.

This SageMaker model package provides a REST api to analyze the sentiment of English sentences.

The API accepts input as JSON, CSV or plain text, and identifies the sentiment (positive or negative) and provides a confidence level (float number from 0 to 1).

Use it to score product reviews and app store feedback, triage support tickets by tone, moderate community content, or track sentiment trends across a review backlog.

We welcome your feedback at [email protected]

Model and training data Version 3 is a ModernBERT-base (answerdotai/ModernBERT-base) text classifier fine-tuned with PyTorch and Hugging Face Transformers on the public IMDB movie review, SST-2 (GLUE), and dair-ai/emotion datasets, and served as ONNX. The model package runs in network isolation on SageMaker, so no data leaves your account. It supports real-time endpoints and batch transform, and takes JSON, CSV, or plain text.

Known limitations - English only. Other languages are not supported. - Binary output (positive or negative). There is no neutral class, so neutral text is forced into the closer of the two labels. - Tuned for short review-style text. Split long documents into sentences or passages before sending.

Measured performance - Accuracy 95.4 percent, macro-F1 0.954, measured on 20,000 held-out examples from the public IMDB, SST-2 (GLUE), and dair-ai/emotion datasets. The split is seeded (70/20/10, seed 42) and was not used in training. - Real-time endpoint on ml.m5.large: 263 ms median per request (275 ms p95). - Throughput on the same instance: about 23 texts per second at a batch size of 32.