Merge pull request #67 from PeterDing/fix/decoding-positions

fix(model_runner): correct position indexing to be 0-based

nanovllm/engine/model_runner.py

@@ -167,7 +167,7 @@ class ModelRunner:
         context_lens = []
         for seq in seqs:
             input_ids.append(seq.last_token)
-            positions.append(len(seq))
+            positions.append(len(seq) - 1)
             context_lens.append(len(seq))
             slot_mapping.append(seq.block_table[-1] * self.block_size + seq.last_block_num_tokens  - 1)
         input_ids = torch.tensor(input_ids, dtype=torch.int64, pin_memory=True).cuda(non_blocking=True)