Fix santacoder.

This commit is contained in:
Nicolas Patry 2023-12-21 15:38:40 +00:00
parent 9f42e5f6fd
commit d3b5ae27b0

View File

@ -71,7 +71,7 @@ def _load_multi_mqa_gptq(
g_idx = weights.get_tensor(f"{prefix}.c_attn.g_idx")
g_idx = g_idx.to(device=weights.device)
bits, groupsize = weights._get_gptq_params()
bits, groupsize, _ = weights._get_gptq_params()
from text_generation_server.utils.layers import HAS_EXLLAMA