Tag: inference
-
Which Model for What: a Hugging Face Model Map for Text, Vision, Audio and Video (Hugging Face Series, Part 17)
A task-to-model map for Hugging Face: which model family to use for chat, search, transcription, speech, captioning, image and video, with sizes, licenses, the right library, and the GPU it needs.
-
Hugging Face Spaces and Gradio: Great for Demos, Wrong for Production (Hugging Face Series, Part 14)
A Hugging Face Space turns a model into a shareable Gradio demo in three files. Here is how a Space runs, a worked example, and the hard line where a demo must graduate to real serving.
-
Text Generation Inference (TGI) in Production: A Real Serving Example (Hugging Face Series, Part 12)
TGI turns a Hugging Face model into an OpenAI-compatible endpoint with one docker run. Here are the flags that decide whether it fits your VRAM, how to consume it, and an honest verdict now that TGI is in maintenance mode and Hugging Face points new builds at vLLM.
-
How to Run a Hugging Face Model: Inference Providers vs Endpoints vs Self-Host (Hugging Face Series, Part 11)
Three ways to serve a Hugging Face model: the serverless Inference Providers proxy, dedicated Inference Endpoints, or self-hosting TGI on your own GPUs. A build-vs-buy verdict for the infra team that owns the bill.
-
Running Your First Model with Transformers Pipelines (Hugging Face Series, Part 4)
The transformers pipeline is the shortest path from a model repo to a running inference call. Here is what actually loads onto your box, and how device, device_map and dtype decide whether a model fits your GPU.
-
NVIDIA Nemotron Foundation Models: Open Weights from Nano to Ultra (NVIDIA AI Series, Part 26)
NVIDIA’s Nemotron family explained: genuinely open weights, data and recipes; the hybrid Mamba-Transformer MoE architecture; Nano, Super and Ultra; and when to self-host open models instead of calling a proprietary API.
-
Inference Economics: Throughput, Latency, Batching and Cost Per Token (NVIDIA AI Series, Part 21)
TTFT, ITL, continuous batching, KV cache pressure, FP8 quantization — this is how you compute and actually drive down $/1M tokens on NVIDIA H100, H200, and B200 GPUs without breaking your latency SLOs.
-
Triton Inference Server vs NIM: When to Use Which (NVIDIA AI Series, Part 19)
Triton Inference Server and NVIDIA NIM solve different problems. This guide breaks down when to use each — and when to run both — covering backends, ensembles, dynamic batching, and the NIM packaged-microservice approach for LLM serving.
-
NVIDIA Dynamo Disaggregated Inference: Prefill, Decode, and KV-Aware Routing at Scale (NVIDIA AI Series, Part 20)
NVIDIA Dynamo separates prefill and decode onto independent GPU pools, routing requests via a KV-aware smart router and transferring KV cache blocks via NIXL. Here is when disaggregation wins, when it does not, and what to validate before committing to the architecture.
-
Deploying and Autoscaling NIM in Production on Kubernetes (NVIDIA AI Series, Part 17)
How to deploy NVIDIA NIM in production using the NIM Operator and Helm, wire autoscaling on the right GPU and KV-cache signals instead of CPU, handle cold-start model load, and run blue-green rollouts without dropping throughput.
-
TensorRT and TensorRT-LLM: Optimization, Quantization, and Engine Building (NVIDIA AI Series, Part 18)
What TensorRT does at build time versus what TensorRT-LLM adds at runtime — kernel fusion, paged KV cache, in-flight batching, and quantization choices from FP8 to NVFP4 — and when to hand-build engines instead of relying on a NIM.
-
NVIDIA NIM Inference Microservices: What a NIM Is and How It Serves a Model (NVIDIA AI Series, Part 16)
NVIDIA NIM packages a model, an optimized inference engine, and an OpenAI-compatible API into a single container. Pull it, pass your NGC API key, and you have a production inference endpoint on your own GPU infrastructure in minutes.
Architect’s Toolkit
PJ’s Tools
VMware Cloud Foundation
- VCF Documentation
- VCF 9 Planning & Preparation Workbook
- VCF Bill of Materials (BoM)
- VMware Compatibility Guide
- VMware Interoperability Matrix
- VMware Configuration Maximums
- VMware Ports & Protocols
- VMware Hands-on Labs
- RVTools Download
Nutanix
AI & Cloud-Native Platform
- NVIDIA Build (Model Catalog)
- NVIDIA AI Enterprise Reference Architecture
- NVIDIA NIM Performance Benchmarking
- NVIDIA NGC Catalog
- NeMo Microservices Helm Chart
- Helm Charts Repository
- Hugging Face Models
Architecture & Design
About the Author
Dr Pranay Jha
Dr. Pranay Jha is a Cloud and AI Consultant with 18+ years of experience in hybrid cloud, virtualization, and enterprise infrastructure transformation. He specializes in VMware technologies, multi-cloud strategy, and Generative AI solutions. He holds a PhD in Computer Applications with research focused on Cloud and AI, has published multiple research papers, and has been a VMware vExpert since 2016 and a VMUG Community Leader.
You May Have Missed
DrJha