Phi 4 FP8 Dynamic

Purpose: Running Phi-4 inference with reduced memory and compute requirements through FP8 quantization.

Inferred · not functionally tested

Audience: machine learning engineers and developers

Inferred · not functionally tested

Functions: Unknown

Interfaces: API: indicated (inferred, not tested) · MCP: unknown · CLI: indicated (inferred, not tested) · Self-hosting: indicated (inferred, not tested)

Recorded constraints: pricing: open_source · license: Apache-2.0 · platforms: WEB · deployment: browser, self_hosted, cli, api_only

Constraint provenance is unknown; confirm requirements with the publisher.

Record sources: huggingface.co. These links do not verify the individual claims.

In the Quantised & converted weights space, Phi 4 FP8 Dynamic takes a focused approach. Inferred · not functionally tested: It focuses on running Phi-4 inference with reduced memory and compute requirements through FP8 quantization. Inferred · not functionally tested: Phi 4 FP8 Dynamic is an open-source project aimed at machine learning engineers and developers. Basis unknown · not verified: The project is open source (Apache-2.0). Basis unknown · not verified: It runs on the web, the command line, and API, and it can be self-hosted.

Behind Phi 4 FP8 Dynamic is Red Hat AI, and it first shipped in 2019. Development happens publicly on GitHub with 3.7k stars and 182 commits in the last 90 days. Inferred · not functionally tested: Key capabilities include FP8 quantization, phi-4 weights, and text generation.

Summary written by a language model from the project’s public pages.