Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Fix model loader for more quantization formats #2448

Merged
merged 1 commit into from
Dec 11, 2024
Merged

Fix model loader for more quantization formats #2448

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
22 changes: 22 additions & 0 deletions python/sglang/srt/models/llama.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -294,6 +294,28 @@ def forward(


class LlamaForCausalLM(nn.Module):

# BitandBytes specific attributes
default_bitsandbytes_target_modules = [
".gate_proj.",
".down_proj.",
".up_proj.",
".q_proj.",
".k_proj.",
".v_proj.",
".o_proj.",
]
# in TP, these weights are partitioned along the column dimension (dim=-1)
column_parallel_weights_modules = [".down_proj.", ".o_proj."]
bitsandbytes_stacked_params_mapping = {
# shard_name, weight_name, index
"q_proj": ("qkv_proj", 0),
"k_proj": ("qkv_proj", 1),
"v_proj": ("qkv_proj", 2),
"gate_proj": ("gate_up_proj", 0),
"up_proj": ("gate_up_proj", 1),
}

def __init__(
self,
config: LlamaConfig,
Expand Down
20 changes: 20 additions & 0 deletions python/sglang/srt/models/qwen2.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -267,6 +267,26 @@ def forward(


class Qwen2ForCausalLM(nn.Module):

# BitandBytes specific attributes
default_bitsandbytes_target_modules = [
".gate_proj.",
".down_proj.",
".up_proj.",
".q_proj.",
".k_proj.",
".v_proj.",
".o_proj.",
]
bitsandbytes_stacked_params_mapping = {
# shard_name, weight_name, index
"q_proj": ("qkv_proj", 0),
"k_proj": ("qkv_proj", 1),
"v_proj": ("qkv_proj", 2),
"gate_proj": ("gate_up_proj", 0),
"up_proj": ("gate_up_proj", 1),
}

def __init__(
self,
config: Qwen2Config,
Expand Down
14 changes: 12 additions & 2 deletions python/sglang/srt/server_args.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -283,7 +283,15 @@ def add_cli_args(parser: argparse.ArgumentParser):
"--load-format",
type=str,
default=ServerArgs.load_format,
choices=["auto", "pt", "safetensors", "npcache", "dummy", "gguf"],
choices=[
"auto",
"pt",
"safetensors",
"npcache",
"dummy",
"gguf",
"bitsandbytes",
],
help="The format of the model weights to load. "
'"auto" will try to load the weights in the safetensors format '
"and fall back to the pytorch bin format if safetensors format "
Expand All @@ -294,7 +302,9 @@ def add_cli_args(parser: argparse.ArgumentParser):
"a numpy cache to speed up the loading. "
'"dummy" will initialize the weights with random values, '
"which is mainly for profiling."
'"gguf" will load the weights in the gguf format. ',
'"gguf" will load the weights in the gguf format. '
'"bitsandbytes" will load the weights using bitsandbytes '
"quantization.",
)
parser.add_argument(
"--trust-remote-code",
Expand Down
Loading