LFM2.5-8B-A1B
PulseAugur coverage of LFM2.5-8B-A1B — every cluster mentioning LFM2.5-8B-A1B across labs, papers, and developer communities, ranked by signal.
1 day(s) with sentiment data
-
Liquid AI boosts LFM2.5 model speed up to 3.18x with DSpark speculative decoding
Liquid AI has released DSpark draft models for its LFM2.5 series, which enhance decoding speed by up to 3.18x without altering output quality. These models utilize speculative decoding, where a smaller draft model propo…
-
LiquidAI upgrades LFM2.5-8B-A1B model with expanded tokenizer
LiquidAI has released LFM2.5-8B-A1B, a model that features an upgraded tokenizer. This enhancement doubles the vocabulary size from 65,000 to 128,000 tokens, specifically addressing issues with fine-grained splitting of…
-
New LLM research covers multimodal alignment, reasoning audits, and energy use · 10 sources tracked
Recent research explores various facets of Large Language Model (LLM) capabilities and limitations. One study investigates alignment in multimodal LLMs, proposing a new data generation method to improve image-text consi…
-
Developer releases Rust-native, CPU-only LFM2.5-8B-A1B implementation
A developer has created a Rust-native, CPU-only implementation of the LFM2.5-8B-A1B language model. This project, still in progress, has been published as a cargo crate and includes features like tool use callbacks. The…
-
Liquid AI ships 8B MoE model with 128K context for on-device tool use
Liquid AI has released LFM2.5-8B-A1B, an 8-billion parameter Mixture-of-Experts model designed for efficient on-device tool calling. This new version significantly expands the context window to 128,000 tokens and scales…
-
LFM2.5 8b A1b model matches Nemotron 3 Nano speed
A new open-source language model, LFM2.5 8b A1b, has been released, offering performance comparable to Nemotron 3 Nano. This model is noted for its high speed and is being integrated into the SmallCode project due to it…
-
LiquidAI releases LFM2.5-8B-A1B model for on-device and server use
LiquidAI has released LFM2.5-8B-A1B, an 8 billion parameter model optimized for on-device applications and efficient server-side use cases. This model boasts a 128K context length and a hybrid MoE architecture, capable …
-
LiquidAI releases on-device LFM2.5-8B-A1B; OpenMOSS-TTS v1.5 adds languages
LiquidAI has released LFM2.5-8B-A1B, a new family of hybrid models optimized for on-device deployment, offering competitive performance with larger models and high inference speeds on both CPU and GPU. Concurrently, the…
-
Liquid AI ships LFM2.5-8B-A1B on-device MoE model
Liquid AI has released LFM2.5-8B-A1B, a new on-device Mixture-of-Experts (MoE) model designed for complex tasks and tool chaining. This model features 8.3 billion total parameters but activates only 1.5 billion per toke…