GitHub - yuanwu2017/llm-dbg

Run test with accelerate on XPU/GPU

accelerate config

CCL_ZE_IPC_EXCHANGE=sockets accelerate launch --main_process_port=29502 ccl_test.py --device xpu --launch torch

accelerate launch --main_process_port=29502  ccl_test.py --device cuda 2>&1 |tee nvidia.log

CCL_ZE_IPC_EXCHANGE=sockets mpirun -n 2 -l  python ccl_test.py --device xpu --launch mpi

Name		Name	Last commit message	Last commit date
Latest commit History 44 Commits
finetune		finetune
speech_vector		speech_vector
README.md		README.md
attn_de.tar.gz.part.aa		attn_de.tar.gz.part.aa
attn_de.tar.gz.part.ab		attn_de.tar.gz.part.ab
attn_de.tar.gz.part.ac		attn_de.tar.gz.part.ac
attn_weight.tar.gz.part.aa		attn_weight.tar.gz.part.aa
attn_weight.tar.gz.part.ab		attn_weight.tar.gz.part.ab
attn_weight.tar.gz.part.ac		attn_weight.tar.gz.part.ac
autogptq_bench.py		autogptq_bench.py
benchmark_inpaint.py		benchmark_inpaint.py
benchmark_tgi.py		benchmark_tgi.py
ccl_test.py		ccl_test.py
compare.py		compare.py
default.txt		default.txt
diffusers_test_output_type.py		diffusers_test_output_type.py
err3.txt		err3.txt
llama4.py		llama4.py
print_model.py		print_model.py
rabbit.jpg		rabbit.jpg
rabbit.png		rabbit.png
requirement.txt		requirement.txt
run_tei_benchmark.sh		run_tei_benchmark.sh
run_text-to-speech.py		run_text-to-speech.py
run_tgi_benchmark.sh		run_tgi_benchmark.sh
run_tgi_llava_next_benchmark.sh		run_tgi_llava_next_benchmark.sh
run_tgi_stream_benchmark.sh		run_tgi_stream_benchmark.sh
tensor.pt		tensor.pt
test_autocast.py		test_autocast.py
test_fft.py		test_fft.py
test_grad_difference.py		test_grad_difference.py
test_reshape.py		test_reshape.py
test_set_queries.tsv		test_set_queries.tsv