Pedro Cuenca's picture

Open to Collab

Pedro Cuenca PRO

pcuenq

huggingface

·

AI & ML interests

None yet

Recent Activity

liked a model about 17 hours ago

unsloth/gemma-4-26B-A4B-it-GGUF

liked a model about 17 hours ago

ggml-org/Qwen3.6-35B-A3B-MTP-GGUF

reacted to alvarobartt's post with 🚀 about 21 hours ago

Latest `hf-mem` release added a breakdown of Mixture-of-Experts (MoE) memory usage! TL; DR MoEs can be misleading to reason about from active parameters alone, since each token only activates a subset of experts, while the serving setup still needs to account for the full resident memory footprint. 🧠 `hf-mem` now splits MoE memory into base model weights, routed experts, and KV cache 🏗️ Dense models usually load and use most weights every forward pass, while MoEs load many experts but only route each token to a few of them ⚡ Active params isn't the same as memory footprint, especially for sparse architectures 📦 Runtime memory is about what is used per request/token, while loading memory also includes the expert weights that need to be resident 📚 KV cache can still dominate depending on context length, batch size, and concurrency 🔀 Expert Parallelism (EP) helps shard experts across accelerators when expert weights dominate 🚀 Data Parallelism (DP) + EP is often a good fit for throughput-oriented MoE serving Check the repository at https://github.com/alvarobartt/hf-mem

View all activity

Organizations

pcuenq 's models 130

pcuenq/gemma-4-E2B-it

Any-to-Any • 5B • Updated 2 days ago • 34

pcuenq/SmolLM3-3B

Text Generation • 3B • Updated 2 days ago • 35

pcuenq/Qwen3.5-9B-mlx

Image-Text-to-Text • 2B • Updated Mar 23 • 50

pcuenq/Qwen3.5-397B-A17B-4bit

Text Generation • 396B • Updated Feb 16 • 181

pcuenq/sam2-hiera-large

Mask Generation • 0.2B • Updated Nov 19, 2025 • 11

pcuenq/candle-depth-anything-v2-small

Updated Nov 10, 2025

pcuenq/nvidia-nano-clone

Image-Text-to-Text • 13B • Updated Nov 5, 2025 • 24

pcuenq/paddle-test-7

Image-Text-to-Text • Updated Oct 25, 2025 • 3

pcuenq/paddle-test-6

Image-Text-to-Text • 1.0B • Updated Oct 25, 2025 • 9

pcuenq/paddle-test-5

Image-Text-to-Text • 1.0B • Updated Oct 25, 2025 • 5

pcuenq/paddle-test-4

Image-Text-to-Text • 1.0B • Updated Oct 25, 2025 • 9

pcuenq/paddle-test-3

Image-Text-to-Text • 1.0B • Updated Oct 25, 2025 • 6

pcuenq/paddle-test-2

Image-Text-to-Text • 1.0B • Updated Oct 25, 2025 • 5

pcuenq/paddle-test-1

Image-Text-to-Text • Updated Oct 25, 2025 • 11

pcuenq/PaddleOCR-VL-0.9B

Image-Text-to-Text • 1.0B • Updated Oct 20, 2025 • 29 • 1

pcuenq/stable-diffusion-v1-5

Text-to-Image • Updated Oct 7, 2025 • 52

pcuenq/sdxl-turbo

Text-to-Image • Updated Oct 7, 2025 • 19

pcuenq/canary-sp-model

Updated Sep 23, 2025

pcuenq/Qwen3-235B-A22B-Instruct-2507-mlx-4Bit

Text Generation • 235B • Updated Aug 14, 2025 • 33 • 1

pcuenq/Phi-4-mini-instruct-mlx-4Bit

Text Generation • 0.6B • Updated Aug 14, 2025 • 48 • 1

pcuenq/Phi-4-mini-instruct-mlx-8Bit

Text Generation • 1B • Updated Aug 14, 2025 • 19

pcuenq/Baichuan-M1-14B-Instruct-tokenizer

Text Generation • Updated Jul 7, 2025 • 5

pcuenq/Hunyuan-7B-Instruct-tokenizer

Text Generation • Updated Jul 7, 2025 • 3

pcuenq/marigold-coco-segmentation

Updated Jul 4, 2025 • 7

pcuenq/t

Updated May 22, 2025

pcuenq/gated-test

Text Generation • Updated Mar 6, 2025

pcuenq/DeepSeek-R1-Distill-Qwen-32B-Q2-6

Text Generation • 4B • Updated Feb 5, 2025 • 36 • 1

pcuenq/jh_dreambooth_1000

Text-to-Image • Updated Feb 5, 2025 • 18 • 1

pcuenq/gemma-2-2b-it-Q4_K_M

Text Generation • 0.4B • Updated Feb 5, 2025 • 62

pcuenq/gpt2-finetuned-cnn-summarization-v2-tokenizer

Text Generation • Updated Jan 31, 2025 • 4