huggingface/text-generation-inference
5
0
Fork 0
mirror of https://github.com/huggingface/text-generation-inference.git synced 2025-09-10 20:04:52 +00:00
Code Issues Packages Projects Releases Wiki Activity

Fixing GTPQ device santacoder.

Browse Source
This commit is contained in:
Nicolas Patry 2023-07-20 19:08:33 +00:00
parent 7faef69015
commit 900ac49454
1 changed files with 5 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
server/text_generation_server/models/custom_modeling/flash_santacoder_modeling.py
View File

@ -49,6 +49,7 @@ def _load_multi_mqa_gptq(
q_tensor = slice_[:, start:stop] q_tensor = slice_[:, start:stop]
kv_tensor = slice_[:, -2 * head_size :] kv_tensor = slice_[:, -2 * head_size :]
qweight = torch.cat([q_tensor, kv_tensor], dim=1) qweight = torch.cat([q_tensor, kv_tensor], dim=1)
qweight = qweight.to(device=weights.device)
slice_ = weights._get_slice(f"{prefix}.c_attn.scales") slice_ = weights._get_slice(f"{prefix}.c_attn.scales")
shape = slice_.get_shape() shape = slice_.get_shape()
@ -59,6 +60,7 @@ def _load_multi_mqa_gptq(
q_tensor = slice_[:, start:stop] q_tensor = slice_[:, start:stop]
kv_tensor = slice_[:, -2 * head_size :] kv_tensor = slice_[:, -2 * head_size :]
scales = torch.cat([q_tensor, kv_tensor], dim=1) scales = torch.cat([q_tensor, kv_tensor], dim=1)
scales = scales.to(device=weights.device)
slice_ = weights._get_slice(f"{prefix}.c_attn.qzeros") slice_ = weights._get_slice(f"{prefix}.c_attn.qzeros")
shape = slice_.get_shape() shape = slice_.get_shape()
@ -69,8 +71,10 @@ def _load_multi_mqa_gptq(
q_tensor = slice_[:, start:stop] q_tensor = slice_[:, start:stop]
kv_tensor = slice_[:, -2 * head_size * 4 // 32 :] kv_tensor = slice_[:, -2 * head_size * 4 // 32 :]
qzeros = torch.cat([q_tensor, kv_tensor], dim=1) qzeros = torch.cat([q_tensor, kv_tensor], dim=1)
qzeros = qzeros.to(device=weights.device)
g_idx = weights.get_tensor(f"{prefix}.c_attn.g_idx") g_idx = weights.get_tensor(f"{prefix}.c_attn.g_idx")
g_idx = g_idx.to(device=weights.device)
bits, groupsize = weights._get_gptq_qparams() bits, groupsize = weights._get_gptq_qparams()
from text_generation_server.utils.layers import HAS_EXLLAMA from text_generation_server.utils.layers import HAS_EXLLAMA
@ -88,6 +92,7 @@ def _load_multi_mqa_gptq(
q_tensor = slice_[start:stop] q_tensor = slice_[start:stop]
kv_tensor = slice_[-2 * head_size :] kv_tensor = slice_[-2 * head_size :]
bias = torch.cat([q_tensor, kv_tensor], dim=0) bias = torch.cat([q_tensor, kv_tensor], dim=0)
bias = bias.to(device=weights.device)
return TensorParallelColumnLinear(get_linear(weight, bias, config.quantize)) return TensorParallelColumnLinear(get_linear(weight, bias, config.quantize))
else: else: