parrishcorcoran
/

MedusaBitNet-2B-4T

Text Generation

speculative-decoding

ternary-weights

efficient-inference

Model card Files Files and versions

MedusaBitNet-2B-4T / figures

1.78 MB

Ctrl+K

Ctrl+K

2 contributors

History: 1 commit

parrishcorcoran's picture

parrishcorcoran

Upload folder using huggingface_hub

0f4a2f7 verified about 2 months ago

architecture.png

163 kB
xet

Upload folder using huggingface_hub about 2 months ago
energy_efficiency.png

136 kB
xet

Upload folder using huggingface_hub about 2 months ago
head_accuracy.png

189 kB
xet

Upload folder using huggingface_hub about 2 months ago
headtohead_throughput.png

119 kB
xet

Upload folder using huggingface_hub about 2 months ago
hero_summary.png

122 kB
xet

Upload folder using huggingface_hub about 2 months ago
medusa_speedup.png

142 kB
xet

Upload folder using huggingface_hub about 2 months ago
memory_vs_energy.png

146 kB
xet

Upload folder using huggingface_hub about 2 months ago
performance_summary.png

153 kB
xet

Upload folder using huggingface_hub about 2 months ago
pipeline_timeline.png

51.9 kB
Upload folder using huggingface_hub about 2 months ago
quality_vs_efficiency.png

148 kB
xet

Upload folder using huggingface_hub about 2 months ago
status_transparency.png

168 kB
xet

Upload folder using huggingface_hub about 2 months ago
throughput_vs_quality.png

152 kB
xet

Upload folder using huggingface_hub about 2 months ago
training_loss.png

85.1 kB
Upload folder using huggingface_hub about 2 months ago