search

BlinkDL / RWKV-LM

RWKV is an RNN with transformer-level LLM performance. It can be directly trained like a GPT (parallelizable). So it's combining the best of RNN and transformer - great performance, fast inference, saves VRAM, fast training, "infinite" ctx_len, and free sentence embedding.
Apache License 2.0
11.99k stars 825 forks source link

可以给个requirements？ #202

Open fuxuelinwudi opened 7 months ago

fuxuelinwudi commented 7 months ago

作者可以给一个 requirements？

BlinkDL commented 7 months ago

deepspeed==0.7.0 pytorch-lightning==1.9.5 torch 1.13.1+cu117

fuxuelinwudi commented 7 months ago

RuntimeError: The following operation failed in the TorchScript interpreter. Traceback of TorchScript (most recent call last): File "/data1/fxl/snn_llm/RWKV-LM-main/RWKV-v4/src/model_origin.py", line 227, in jit_func

    # Mix x with the previous timestep to produce xk, xv, xr
    xx = self.time_shift(x)
         ~~~~~~~~~~~~~~~ <--- HERE
    xk = x * self.time_mix_k + xx * (1 - self.time_mix_k)
    xv = x * self.time_mix_v + xx * (1 - self.time_mix_v)

File "/home/ubuntu/miniconda3/envs/cqw/lib/python3.9/site-packages/torch/nn/modules/padding.py", line 25, in forward def forward(self, input: Tensor) -> Tensor: return F.pad(input, self.padding, 'constant', self.value)


RuntimeError: CUDA error: device-side assert triggered
CUDA kernel errors might be asynchronously reported at some other API call,so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1.
humanpp commented 7 months ago

我试的能运行的， pytorch直接更新到最新版 pip install --upgrade torch pytorch-lightning==1.9.2就行 deepspeed==0.7.0如果出问题的话，也更新到最新版 pip install --upgrade deepspeed

Yang-125 commented 7 months ago

rwkv-4 最低版本的torch有限制吗？