google

/

timesfm-2.5-200m-pytorch

like 123

Follow

Google 32.8k

Hey I am testing TimesFM for my project, everything was working fine until today. I didn't change the code, in fact it is deafult code from your example

model = timesfm.TimesFM_2p5_200M_torch.from_pretrained("google/timesfm-2.5-200m-pytorch", torch_compile=True)

RuntimeError: Error(s) in loading state_dict for TimesFM_2p5_200M_torch_module:
    Missing key(s) in state_dict: "stacked_xf.0.attn.query.weight", "stacked_xf.0.attn.key.weight", "stacked_xf.0.attn.value.weight", "stacked_xf.1.attn.query.weight", "stacked_xf.1.attn.key.weight", "stacked_xf.1.attn.value.weight", "stacked_xf.2.attn.query.weight", "stacked_xf.2.attn.key.weight", "stacked_xf.2.attn.value.weight", "stacked_xf.3.attn.query.weight", "stacked_xf.3.attn.key.weight", "stacked_xf.3.attn.value.weight", "stacked_xf.4.attn.query.weight", "stacked_xf.4.attn.key.weight", "stacked_xf.4.attn.value.weight", "stacked_xf.5.attn.query.weight", "stacked_xf.5.attn.key.weight", "stacked_xf.5.attn.value.weight", "stacked_xf.6.attn.query.weight", "stacked_xf.6.attn.key.weight", "stacked_xf.6.attn.value.weight", "stacked_xf.7.attn.query.weight", "stacked_xf.7.attn.key.weight", "stacked_xf.7.attn.value.weight", "stacked_xf.8.attn.query.weight", "stacked_xf.8.attn.key.weight", "stacked_xf.8.attn.value.weight", "stacked_xf.9.attn.query.weight", "stacked_xf.9.attn.key.weight", "stacked_xf.9.attn.value.weight", "stacked_xf.10.attn.query.weight", "stacked_xf.10.attn.key.weight", "stacked_xf.10.attn.value.weight", "stacked_xf.11.attn.query.weight", "stacked_xf.11.attn.key.weight", "stacked_xf.11.attn.value.weight", "stacked_xf.12.attn.query.weight", "stacked_xf.12.attn.key.weight", "stacked_xf.12.attn.value.weight", "stacked_xf.13.attn.query.weight", "stacked_xf.13.attn.key.weight", "stacked_xf.13.attn.value.weight", "stacked_xf.14.attn.query.weight", "stacked_xf.14.attn.key.weight", "stacked_xf.14.attn.value.weight", "stacked_xf.15.attn.query.weight", "stacked_xf.15.attn.key.weight", "stacked_xf.15.attn.value.weight", "stacked_xf.16.attn.query.weight", "stacked_xf.16.attn.key.weight", "stacked_xf.16.attn.value.weight", "stacked_xf.17.attn.query.weight", "stacked_xf.17.attn.key.weight", "stacked_xf.17.attn.value.weight", "stacked_xf.18.attn.query.weight", "stacked_xf.18.attn.key.weight", "stacked_xf.18.attn.value.weight", "stacked_xf.19.attn.query.weight", "stacked_xf.19.attn.key.weight", "stacked_xf.19.attn.value.weight". 
    Unexpected key(s) in state_dict: "stacked_xf.0.attn.qkv_proj.weight", "stacked_xf.1.attn.qkv_proj.weight", "stacked_xf.2.attn.qkv_proj.weight", "stacked_xf.3.attn.qkv_proj.weight", "stacked_xf.4.attn.qkv_proj.weight", "stacked_xf.5.attn.qkv_proj.weight", "stacked_xf.6.attn.qkv_proj.weight", "stacked_xf.7.attn.qkv_proj.weight", "stacked_xf.8.attn.qkv_proj.weight", "stacked_xf.9.attn.qkv_proj.weight", "stacked_xf.10.attn.qkv_proj.weight", "stacked_xf.11.attn.qkv_proj.weight", "stacked_xf.12.attn.qkv_proj.weight", "stacked_xf.13.attn.qkv_proj.weight", "stacked_xf.14.attn.qkv_proj.weight", "stacked_xf.15.attn.qkv_proj.weight", "stacked_xf.16.attn.qkv_proj.weight", "stacked_xf.17.attn.qkv_proj.weight", "stacked_xf.18.attn.qkv_proj.weight", "stacked_xf.19.attn.qkv_proj.weight". 

This version is also not working

model = timesfm.TimesFM_2p5_200M_torch.from_pretrained("google/timesfm-2.5-200m-pytorch")

I reinstalled the package and it's working correctly now.