Add archive notice. (#37)

Added support for dark mode (#23)

Experiment with quantize to FP16 + more perf benchmarks

[HUGE] DistilBERT demo app + cleanup 🎉 (#13)

[HUGE] 🐎 DistilGPT-2, DistilBERT-squad

Update LayerNorm attribute in gpt2 conversion script

Should Fix #9

Fix for error “Bias layer '0_block_attn_afterbias' cannot be 2 dimensional. Must be 1D or 3D.”

TPS

[GPT-2] Speed benchmark

[distilbert] More perf benchmark

[distilbert] Automated conversion: pytorch -> onnx -> coreml

[distilbert] Preliminary perf benchmarks

Added model with 64 token context that can run on iPhone X.

Create LICENSE