List your company

Language Modeling on WikiText-103 (PPL and FLOPs)

5.47Perplexity (PPL)LongLoRA-7B-32k
4.73529.695114.65519.6149Mar 19, 2025May 26, 2025Aug 3, 2025Oct 11, 2025Dec 19, 2025Feb 26, 2026May 6, 2026

Evaluation results

MethodLinks
LongLoRA-7B-32kModel Size=7B, Context Length=16k2025.035.47—
LLaMA-2-7BModel Size=7B, Context Length=4k2025.035.66—
YARN-128k-7bModel Size=7B, Context Length=1k2025.035.71—
YARN-128k-7bModel Size=7B, Context Length=4k2025.035.71—
YARN-128k-7bModel Size=7B, Context Length=2k2025.036.11—
LLaMA-2-7BModel Size=7B, Context Length=2k2025.036.49—
Phi3-128kModel Size=3B, Context Length=4k2025.037.01—
Phi3-128kModel Size=3B, Context Length=16k, GPU Configuration=Dual GPU2025.037.2—
Phi3-128kModel Size=3B, Context Length=2k2025.037.22—
Phi3-128kModel Size=3B, Context Length=1k2025.037.54—
ERMAR (w/ 4K capacity)Model Size=3B, Context Length=2k, Memory Capacity=4K2025.037.61—
ERMAR (w/ 4K capacity)Model Size=3B, Context Length=4k, Memory Capacity=4K2025.037.62—
ERMAR (w/ 4K capacity)Model Size=3B, Context Length=16k, Memory Capacity=4K2025.037.8—
LongLoRA-7B-32kModel Size=7B, Context Length=2k2025.037.83—
LongLLaMA-3B†Model Size=3B, Context Length=4k2025.037.84—
MemLong (w/ 4K capacity)Model Size=3B, Context Length=4k, Memory Capacity=4K2025.037.87—
MemLong (w/ 4K capacity)Model Size=3B, Context Length=2k, Memory Capacity=4K2025.037.92—
MemLong-3B*Model Size=3B, Context Length=2k2025.037.93—
LongLoRA-7B-32kModel Size=7B, Context Length=1k2025.037.99—
MemLong (w/ 4K capacity)Model Size=3B, Context Length=16k, Memory Capacity=4K2025.037.99—
LongLLaMA-3B*Model Size=3B, Context Length=2k2025.038.07—
OpenLLaMA-3BModel Size=3B, Context Length=2k2025.038.08—
LongLLaMA-3B†Model Size=3B, Context Length=2k2025.038.33—
LongLoRA-7B-32kModel Size=7B, Context Length=4k2025.038.39—
ERMAR (w/ 4K capacity)Model Size=3B, Context Length=1k, Memory Capacity=4K2025.038.42—
MemLong (w/ 4K capacity)Model Size=3B, Context Length=1k, Memory Capacity=4K2025.038.53—
MemLong-3B*Model Size=3B, Context Length=1k2025.038.72—
LongLLaMA-3B*Model Size=3B, Context Length=1k2025.038.88—
OpenLLaMA-3BModel Size=3B, Context Length=4k2025.0310—
LongLLaMA-3B*Model Size=3B, Context Length=4k2025.0310—
MemLong-3B*Model Size=3B, Context Length=4k2025.0310—
OpenLLaMA-3BModel Size=3B, Context Length=1k2025.0310.57—
LongLLaMA-3B†Model Size=3B, Context Length=1k2025.0310.69—
LLaMA-2-7BModel Size=7B, Context Length=1k2025.0310.82—
AIR-MoESize=Medium2026.0518.62755.9
PEERSize=Medium2026.0518.71751
Coarse MoESize=Medium2026.0518.79836.7
Hierarchical MoESize=Medium2026.0519.27954.7
Standard Granular MoESize=Small2026.0521.34467.5
AIR-MoESize=Small2026.0521.82324.1
PEERSize=Small2026.0522.25352.8
Hierarchical MoESize=Small2026.0522.55350.4
Coarse MoESize=Small2026.0523.84342.5