8000
We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
There was an error while loading. Please reload this page.
1 parent 52350cc commit d9b38e3Copy full SHA for d9b38e3
llama_cpp/llama.py
@@ -1019,12 +1019,11 @@ def eval(self, tokens: Sequence[int]):
1019
"""
1020
assert self._ctx.ctx is not None
1021
assert self._batch.batch is not None
1022
- n_ctx = self._n_ctx
+ self._ctx.kv_cache_seq_rm(-1, self.n_tokens, -1)
1023
for i in range(0, len(tokens), self.n_batch):
1024
batch = tokens[i : min(len(tokens), i + self.n_batch)]
1025
- n_past = min(n_ctx - len(batch), self.n_tokens)
+ n_past = self.n_tokens
1026
n_tokens = len(batch)
1027
- self._ctx.kv_cache_seq_rm(-1, n_past, -1)
1028
self._batch.set_batch(
1029
batch=batch, n_past=n_past, logits_all=self.context_params.logits_all
1030
)
0 commit comments