phoenix-oss/llama-stack-mirror
0
0
Fork 1
mirror of https://github.com/meta-llama/llama-stack.git synced 2025-10-04 04:04:14 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

updating license for toolchain

Browse source
This commit is contained in:
Ashwin Bharambe 2024-07-22 20:31:42 -07:00
parent 0e2fc9966a
commit 86fff23a9e
74 changed files with 512 additions and 94 deletions
Download patch file Download diff file Expand all files Collapse all files

39
llama_toolchain/cli/inference/configure.py
View file

@ -1,10 +1,17 @@
# Copyright (c) Meta Platforms, Inc. and affiliates.
# All rights reserved.
#
# This source code is licensed under the terms described found in the
# LICENSE file in the root directory of this source tree.
import argparse
import os
import pkg_resources
import textwrap
from pathlib import Path
import pkg_resources
from llama_toolchain.cli.subcommand import Subcommand
from llama_toolchain.utils import DEFAULT_DUMP_DIR
@ -36,21 +43,22 @@ class InferenceConfigure(Subcommand):
pass
def read_user_inputs(self):
checkpoint_dir = input("Enter the checkpoint directory for the model (e.g., ~/.llama/checkpoints/Meta-Llama-3-8B/): ")
model_parallel_size = input("Enter model parallel size (e.g., 1 for 8B / 8 for 70B and 405B): ")
assert model_parallel_size.isdigit() and int(model_parallel_size) in {1, 8}, "model parallel size must be 1 or 8"
checkpoint_dir = input(
"Enter the checkpoint directory for the model (e.g., ~/.llama/checkpoints/Meta-Llama-3-8B/): "
)
model_parallel_size = input(
"Enter model parallel size (e.g., 1 for 8B / 8 for 70B and 405B): "
)
assert model_parallel_size.isdigit() and int(model_parallel_size) in {
1,
8,
}, "model parallel size must be 1 or 8"
return checkpoint_dir, model_parallel_size
def write_output_yaml(
self,
checkpoint_dir,
model_parallel_size,
yaml_output_path
):
def write_output_yaml(self, checkpoint_dir, model_parallel_size, yaml_output_path):
default_conf_path = pkg_resources.resource_filename(
'llama_toolchain',
'data/default_inference_config.yaml'
"llama_toolchain", "data/default_inference_config.yaml"
)
with open(default_conf_path, "r") as f:
yaml_content = f.read()
@ -60,7 +68,7 @@ class InferenceConfigure(Subcommand):
model_parallel_size=model_parallel_size,
)
with open(yaml_output_path, 'w') as yaml_file:
with open(yaml_output_path, "w") as yaml_file:
yaml_file.write(yaml_content.strip())
print(f"YAML configuration has been written to {yaml_output_path}")
@ -69,8 +77,9 @@ class InferenceConfigure(Subcommand):
checkpoint_dir, model_parallel_size = self.read_user_inputs()
checkpoint_dir = os.path.expanduser(checkpoint_dir)
assert Path(checkpoint_dir).exists() and Path(checkpoint_dir).is_dir(), \
f"{checkpoint_dir} does not exist or it not a directory"
assert (
Path(checkpoint_dir).exists() and Path(checkpoint_dir).is_dir()
), f"{checkpoint_dir} does not exist or it not a directory"
os.makedirs(CONFIGS_BASE_DIR, exist_ok=True)
yaml_output_path = Path(CONFIGS_BASE_DIR) / "inference.yaml"