|
--- |
|
language: |
|
- en |
|
tags: |
|
- pytorch |
|
- text-generation |
|
- causal-lm |
|
- rwkv |
|
license: apache-2.0 |
|
datasets: |
|
- The Pile |
|
|
|
--- |
|
|
|
# RWKV-2 430M |
|
|
|
## Model Description |
|
|
|
RWKV-2 430M is a L24-D1024 causal language model trained on the Pile. See https://github.com/BlinkDL/RWKV-LM for details. |
|
|
|
At this moment you have to use my Github code (https://github.com/BlinkDL/RWKV-v2-RNN-Pile) to run it. |
|
|
|
ctx_len = 768 n_layer = 24 n_embd = 1024 |
|
|
|
Final checkpoint: 20220615-10803.pth : Trained on the Pile for 331B tokens. |
|
* Pile loss 2.349 |
|
* LAMBADA ppl 15.34, acc 42.42% |
|
* PIQA acc 67.03% |
|
* SC2016 acc 62.05% |
|
* Hellaswag acc_norm 38.47% |
