---
language:
- en
tags:
- pytorch
- text-generation
- causal-lm
- rwkv
license: apache-2.0
datasets:
- The Pile

---

# RWKV-4 430M

## Model Description

RWKV-4 430M is a L24-D1024 causal language model trained on the Pile. See https://github.com/BlinkDL/RWKV-LM for details.

** Note: It's a BF16 model, and it may overflow if you are using FP16 (probably fixable by rescaling the weights). **

At this moment you have to use my Github code (https://github.com/BlinkDL/RWKV-LM) to run it.

ctx_len = 1024
n_layer = 24
n_embd = 1024

Final checkpoint:
RWKV-4-Pile-430M-20220808-8066.pth : Trained on the Pile for 333B tokens.
* Pile loss 2.2621
* LAMBADA ppl 13.04, acc 45.16%
* PIQA acc 67.52%
* SC2016 acc 63.87%
* Hellaswag acc_norm 40.90%