|
pulsatrix 1.1.0
|
A right-padded training batch. More...
#include <lm_scoring.hpp>

Public Attributes | |
| Tensor | ids |
| (N, L) token ids as floats: prompt, then completion but its last token | |
| Tensor | targets |
| (N, L): the next token where it is a completion token, else kIgnoreIndex | |
| std::vector< int64_t > | lengths |
| each row's real length, |prompt| + |completion| - 1 | |
| int64_t | target_tokens = 0 |
| completion tokens in the batch, the loss's token count | |
A right-padded training batch.
| Tensor pulsatrix::CausalLmBatch::ids |
(N, L) token ids as floats: prompt, then completion but its last token
| std::vector<int64_t> pulsatrix::CausalLmBatch::lengths |
each row's real length, |prompt| + |completion| - 1
| int64_t pulsatrix::CausalLmBatch::target_tokens = 0 |
completion tokens in the batch, the loss's token count
| Tensor pulsatrix::CausalLmBatch::targets |
(N, L): the next token where it is a completion token, else kIgnoreIndex