Find answers from the community

Home
Members
OceanLi
O
OceanLi
Offline, last seen 3 months ago
Joined September 25, 2024
Running the REBEL + wiki KG index, somehow pytorch error?
Plain Text
File ~/anaconda3/envs/test_cluster2/lib/python3.11/site-packages/transformers/generation/beam_search.py:397, in BeamSearchScorer.finalize(self, input_ids, final_beam_scores, final_beam_tokens, final_beam_indices, max_length, pad_token_id, eos_token_id, beam_indices, decoder_prompt_len)
    395 # fill with hypotheses and eos_token_id if the latter fits in
    396 for i, (hypo, best_idx) in enumerate(zip(best, best_indices)):
--> 397     decoded[i, : sent_lengths[i]] = hypo
    399     if indices is not None:
    400         indices[i, : len(best_idx)] = torch.tensor(best_idx)

RuntimeError: The expanded size of the tensor (200) must match the existing size (38) at non-singleton dimension 0.  Target sizes: [200].  Tensor sizes: [38]
83 comments
O
L
d
O
OceanLi
·

Thread

2 comments
O
L
how can I do code splitter with recurisve retrieval ?
5 comments
O
L
@Logan M Hi, I wrote a stock_price_query_engine, feel like it looks interesting, is it worthy of putting onto doc as a tutorial?
11 comments
O
L
@Logan M circular import?
15 comments
O
b
L
hello, I am new into llama_index, and am quite enthusiastic about LLM, I am especially interested in knowledge graph, and its potential in improving retrieval, I would like to contribute to the community, where should I start?
100 comments
O
L
N