LFM2.5-350M
PulseAugur coverage of LFM2.5-350M — every cluster mentioning LFM2.5-350M across labs, papers, and developer communities, ranked by signal.
2 day(s) with sentiment data
-
Hugging Face details fine-tuning for structured LLM outputs
Hugging Face has detailed a method for fine-tuning a 350 million parameter model to improve its ability to generate structured outputs. This process, which involves 100 GRPO steps, aims to enhance schema compliance, a c…
-
Liquid AI releases QAD versions of LFM2.5 models for efficient edge deployment
Liquid AI has released a Quantization-Aware Distillation (QAD) version of its LFM2.5 series of small AI models. This new version is designed to reduce memory usage while minimizing performance degradation, making the mo…
-
Budget PC LLM Test: LFM2.5-1.2B-Instruct Wins for General Use
A developer tested five small LLMs (under 2 billion parameters) on a standard office PC without a dedicated GPU to determine which models perform best on budget hardware. The tests focused on token-per-second speed and …
-
SupraLabs releases expanded context Supra-1.5 and specialized Supra-Title models
SupraLabs has released two new experimental models: the Supra-1.5-exp family, which includes a base model and an instruct-tuned version with a 5x larger context window (5,120 tokens) and expanded training data, and Supr…
-
Jetson Orin Nano benchmarks 8 tiny LLMs across power modes
A benchmark of eight small language models (135M to ~1B parameters) was conducted on a Jetson Orin Nano Super 8GB device. The tests explored four power modes (7W, 15W, 25W, MAXN) using the llama.cpp CUDA backend. The fi…