--- license: apache-2.0 datasets: - togethercomputer/RedPajama-Data-1T-Sample language: - en tags: - llama - llama 2 --- # TinyLlama-1.1B-32k 32k context finetune of TinyLlama-1.1B using increased rope theta (rope frequency base) meant to serve as a long-context speculative decoding model. Created using [TinyLlama-1.1B](https://huggingface.co./TinyLlama/tinyLlama-intermediate-checkpoints-after-1T-token) and further pretraining at 32768 context length on [togethercomputer/RedPajama-Data-1T-Sample](https://huggingface.co./datasets/togethercomputer/RedPajama-Data-1T-Sample). Of note, the base checkpoint used was from commit "final model" fad4f1a5cd0563ac41349b8fec2e6e51156568a0 which was subsequently reverted, and not the current main branch 3T checkpoint of TinyLlama-1.1B. [EXL2 Quants by turboderp](https://huggingface.co./turboderp/TinyLlama-1B-32k-exl2) The quantized model fits alongside a 4.25bpw 70B model at 32k sequence length on a single A6000 and provides noticeable speed-up with speculative decoding. ### Wikitext (wikitext-2-raw-v1_train) Perplexity (64 rows) as evaluated via [exllamav2](https://github.com/turboderp/exllamav2): | Model | 2048 | 4096 | 8192 | 16384 | 32768 | | ---------------------- | ---------- | ---------- | ---------- | ---------- | ---------- | | TinyLlama-1.1B | **8.5633** | 208.3586 | 863.7507 | 1600.5021 | 6981.9021 | | **TinyLlama-1.1B-32k** | 8.6548 | **7.8339** | **7.4904** | **7.3674** | **7.1338** | ### Evaluation on HumanEval by [turboderp](https://huggingface.co./turboderp): | Model | Pass@1 | Pass@10 | | -------------------------------------- | --------------- | ----------- | | TinyLlama-1.1B | **0.0841** | **0.1524** | | TinyLlama-1.1B (NTK alpha=7.7) | 0.0598 | 0.1098 | | TinyLlama-1.1B-32k-ckpt-554 | 0.0732 | 0.1402 | | **TinyLlama-1.1B-32k** | 0.0829 | **0.1524** |