Was trying the model . getting errors . May be I am doing something wrong. Please share complete code.
inputs = tokenizer(["Tell me about Obama"], return_tensors="pt")
inputs=inputs.to(device='cuda:0')
Example 1: Print the scores for each token generated with Greedy Search
outputs = model.generate(**inputs, max_new_tokens=100, return_dict_in_generate=True, output_scores=True)
Results in error
RuntimeError: Triton Error [CUDA]: invalid argument
Hi, could you share the output of pip list? likely you have an incompatible version of something
inputs = tokenizer(["Tell me about Obama"], return_tensors="pt")
inputs=inputs.to(device='cuda:0')Example 1: Print the scores for each token generated with Greedy Search
outputs = model.generate(**inputs, max_new_tokens=100, return_dict_in_generate=True, output_scores=True)
Results in error
RuntimeError: Triton Error [CUDA]: invalid argument
Here's a small colab that works: https://colab.research.google.com/drive/19YGJ-eDe2Wm17hc9hLwobckHZjFB8lo5?usp=sharing
inputs = tokenizer(["Tell me about Obama"], return_tensors="pt")
inputs=inputs.to(device='cuda:0')Example 1: Print the scores for each token generated with Greedy Search
outputs = model.generate(**inputs, max_new_tokens=100, return_dict_in_generate=True, output_scores=True)
Results in error
RuntimeError: Triton Error [CUDA]: invalid argumentHere's a small colab that works: https://colab.research.google.com/drive/19YGJ-eDe2Wm17hc9hLwobckHZjFB8lo5?usp=sharing
Thanks. This works.