Repository Issue Activity (beta)

#338 Can RWKV state encode abstract behavioral dispositions? | RWKV state 能否编码抽象行为倾向？

#349 Structured experiments on G1h 2.9B: L_state trajectory reg, N/K/mode effort frontier, causal state probe, multi-slot capacity

#343 RWKV-v7/rwkv_v7_numpy.py fails on NumPy <2.0 (`.mT` attribute)

#347 67

#346 Out-of-range token IDs trigger a CUDA device-side assert instead of a validation error

#339 Рэпчина

#337 Add support for true unlimited context

#332 RWKV_JIT_ON=0` is ignored when `RWKV_V7_ON=1

#335 Solution to the gating issue in RNN models.

#331 Actual prompts? Old translation promt not working

#330 token填充问题

#329 Is this website, `www.rwkv.cn`, yours?

#149 报错：assert fragment_start < fragment_end

#326 A cleaner Read Me file

#324 Adding support for batch processing

#317 log-linear-attention colab?

#323 Add support for pytorch for both training and interference

#328 Please add support for layer reusing

#325 Add support for sparse layer and dense layer control

#318 Gives wrong output under very long context in LM Studio

#322 Adding support for other tokenizer

#321 Please add support for huggingface dataloader class for training data by streaming

#319 Architecture history link is a 404

#309 Question about the loss backward with RWKV7

#312 Possible bug in `__setstate__`: missing `skip_warmup` parameter in binidx.py