Accelerating vision-language models with LFM2.5-VL-DSpark
As with our recently released LFM2.5-DSpark drafter models, it adds a speculative decoding path that trades a minimal increase in memory footprint for a larger speedup without changing output quality. Faster inference: decode speedups up to 3.13x on device and 2.66x on an H100, with end-to-end gains up to 2.62x and 2.27x. Image patches and text tokens are projected into a shared representation before those layers, so the drafter operates on hidden-state vectors of identical dimensionality regardless of input modality.
- ▪As with our recently released LFM2.5-DSpark drafter models, it adds a speculative decoding path that trades a minimal increase in memory footprint for a larger speedup without changing output quality.
- ▪Faster inference: decode speedups up to 3.13x on device and 2.66x on an H100, with end-to-end gains up to 2.62x and 2.27x.
- ▪Image patches and text tokens are projected into a shared representation before those layers, so the drafter operates on hidden-state vectors of identical dimensionality regardless of input modality.
Hugging Face Blog files mainly under ai. We currently carry 35 of its stories.
Story provenance
Source · retrieval · rights · ranking — open for full record
inspect →
Story provenance
Attribution is not the same as permission. This drawer separates discovery metadata, excerpts, WeSearch-generated summaries, reuse status, and whether the publisher receives the visit. Nothing here claims a legal grant the publisher has not made.
Record
| Original publisher | Hugging Face Blog |
| Canonical URL | https://huggingface.co/blog/LiquidAI/lfm2-5-vl-dspark |
| Publication time | Thu, 24 Sep 2026 14:08:57 GMT |
| Retrieval time | 2026-09-24T14:10:25.985Z |
| Last seen | 2026-09-24T14:10:25.985Z |
| Headline source | Publisher (no WeSearch rewrite) |
| Excerpt source | publisher body |
| Excerpt method | First ~120 words (~800 chars) of extracted publisher body, fair-use limited. |
| Summary | WeSearch · cerebras-chat (WeSearch summarizer) |
| Summary source text | contentText |
| Citation coverage | Summary is a WeSearch-generated derivative; primary citation is the original publisher URL. |
| Cluster | L9RuZR5-Sss2 · 1 stories |
| Cluster logic | Grouped by semantic title/content similarity across sources within a rolling window. Same-publisher template collisions are excluded from coverage comparison. |
| Ranking reason | Story pages are not engagement-ranked. Hub feeds use recency, with optional source-diversified chronological ordering (cap consecutive stories per source). No personalized ranking. |
| Publisher visit | Yes — open original |
| Substitutes article? | No — link-out required for full text |
Rights status (four layers)
WeSearch handling by dimension
| Indexing | May the item be indexed (stored, ranked, made findable)? | Allowed |
| Snippet | May a short excerpt of the publisher's text be shown? | Allowed |
| AI summary | May WeSearch generate its own short summary of the article? | Limited |
| Retrieval / RAG | May the content be exposed for third-party retrieval-augmented generation? | Not asserted |
| Model training | May the content be used to train AI models? | Not asserted |
| Commercial reuse | May the content be reused commercially? | Not permitted |
Basis: Derived from the published RSS/Atom feed. Contact: [email protected]. Reviewed: 2026-07-24.
Opening excerpt (first ~120 words) tap to expand
Back to Articles Accelerating vision-language models with LFM2.5-VL-DSpark Team Article Published September 24, 2026 Upvote - xx tugot17 Follow LiquidAI Yuri Khrustalev ykhrustalev Follow LiquidAI Leonie Monigatti iamleonie Follow LiquidAI Viviana Márquez vivianamarquez Follow LiquidAI How does speculative decoding work for VLMs Training and Architecture Inference Speedup on CPU and GPU Limitations of speculation for vision workloads How to use LFM2.5-VL-DSpark Get Started Citation Today, we release an experimental DSpark draft model for our vision-language model (VLM) LFM2.5-VL-3B. As with our recently released LFM2.5-DSpark drafter models, it adds a speculative decoding path that trades a minimal increase in memory footprint for a larger speedup without changing output quality.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at Hugging Face Blog.
Discussion
0 commentsMore from Hugging Face Blog
