Home/Events/Release of LFM2.5 Q4_0 Checkpoints from Quantization-Aware Distillation

Release of LFM2.5 Q4_0 Checkpoints from Quantization-Aware Distillation

Confirmed

Confidence

90%

Impact: 80%

Updated 2h ago

Consensus Brief

Hugging Face has released updated 4-bit checkpoints for the LFM2.5 models, which include LFM2.5-230M, LFM2.5-350M, LFM2.5-1.2B-Instruct, and LFM2.5-2.6B. These checkpoints utilize Quantization-Aware Distillation (QAD) to maintain performance while reducing memory usage and increasing throughput.

Sourced from

Hugging FaceT1

Primary: Hugging Face

What Changed Since Last Update

2h ago

The new QAD Q4_0 checkpoints recover 97% of the average accuracy lost to quantization compared to previous models.

Claim Ledger

3 claims tracked across sources

Confirmed Fact

QAD checkpoints retain 97.1%, 96.5%, 97.4%, and 96.6% of their respective BF16 baseline performance.

Confirmed Fact

The QAD Q4_0 checkpoints match Q5_K_M quality with a 4-33% higher decode throughput.

Official Claim

The QAD Q4_0 checkpoints are available on Hugging Face.

Role-Based Impact Analysis

Source Timeline

1 source corroborating

LFM2.5 Q4\_0 Checkpoints from Quantization-Aware Distillation

Hugging Face·2h ago