Running bench_qwen.sh in Docker build doesn't work #40
Description
I tried running the bench_qwen.sh example script, and it doesn't work at all. It spits out a bunch of errors, and I'm not sure what to do with it.
According to the Performance Benchmark --> Quick Start heading (README link following), I should be able to run that script with no parameters.
https://github.com/NVIDIA/TileGym/blob/main/modeling/transformers/README.md
- Windows 11
- NVIDIA GeForce RTX 5070 Ti 16 GB
- Docker Desktop latest v4.57.0
- NVIDIA GeForce driver latest 591.74
(TileGym) PS C:\git\TileGym> docker run --gpus all -it tilegym-transformers bash
==========
== CUDA ==
==========
CUDA Version 13.1.0
Container image Copyright (c) 2016-2023, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
This container image and its contents are governed by the NVIDIA Deep Learning Container License.
By pulling and using the container, you accept the terms and conditions of this license:
https://developer.nvidia.com/ngc/nvidia-deep-learning-container-license
A copy of this license is made available in this container at /NGC-DL-CONTAINER-LICENSE for your convenience.
root@3fbe9ab5b1b4:/workspace/tilegym/modeling/transformers# bash bench_qwen.sh
bench_qwen.sh: line 4: $'\r': command not found
bench_qwen.sh: line 7: $'\r': command not found
: invalid optionine 8: set: -
set: usage: set [-abefhkmnptuvxBCHP] [-o option-name] [--] [arg ...]
bench_qwen.sh: line 9: $'\r': command not found
bench_qwen.sh: line 15: $'\r': command not found
========================================
Qwen2-7B Performance Benchmark
========================================
Model: Qwen/Qwen2-7B
Input: sample_inputs/input_prompt_small.txt
tokenslength: 50
Batch size: 16
bench_qwen.sh: line 25: $'\r': command not found
bench_qwen.sh: line 28: $'\r': command not found
Running PyTorch baseline...
usage: infer.py [-h] [--use_tilegym] [--model_id MODEL_ID] [--batch_size BATCH_SIZE] [--input_text INPUT_TEXT] [--num_runs NUM_RUNS] [--warmup_runs WARMUP_RUNS] [--show_outputs] [--summary_file SUMMARY_FILE] [--use_attn] [--use_cutile] [--profile] [--log_dir LOG_DIR]
[--precision PRECISION] [--sentence_file SENTENCE_FILE] [--output_length OUTPUT_LENGTH] [--mock_input_len MOCK_INPUT_LEN]
infer.py: error: unrecognized arguments:
bench_qwen.sh: line 31: --model_id: command not found
bench_qwen.sh: line 32: --profile: command not found
bench_qwen.sh: line 33: --sentence_file: command not found
bench_qwen.sh: line 34: --batch_size: command not found
bench_qwen.sh: line 35: --output_length: command not found
bench_qwen.sh: line 36: --summary_file: command not found
bench_qwen.sh: line 37: $'\r': command not found
Running TileGym CUTILE backend...
usage: infer.py [-h] [--use_tilegym] [--model_id MODEL_ID] [--batch_size BATCH_SIZE] [--input_text INPUT_TEXT] [--num_runs NUM_RUNS] [--warmup_runs WARMUP_RUNS] [--show_outputs] [--summary_file SUMMARY_FILE] [--use_attn] [--use_cutile] [--profile] [--log_dir LOG_DIR]
[--precision PRECISION] [--sentence_file SENTENCE_FILE] [--output_length OUTPUT_LENGTH] [--mock_input_len MOCK_INPUT_LEN]
infer.py: error: unrecognized arguments:
bench_qwen.sh: line 41: --model_id: command not found
bench_qwen.sh: line 42: --use_tilegym: command not found
bench_qwen.sh: line 43: --use_cutile: command not found
bench_qwen.sh: line 44: --use_attn: command not found
bench_qwen.sh: line 45: --profile: command not found
bench_qwen.sh: line 46: --sentence_file: command not found
bench_qwen.sh: line 47: --batch_size: command not found
bench_qwen.sh: line 48: --output_length: command not found
bench_qwen.sh: line 49: --summary_file: command not found
bench_qwen.sh: line 50: $'\r': command not found
========================================
Benchmark Results
========================================
bench_qwen.sh: line 62: syntax error: unexpected end of file
root@3fbe9ab5b1b4:/workspace/tilegym/modeling/transformers#