Skip to content

Commit f383c8f

Browse files
bvandermoonbvandermoon
authored
Merge pull request #58 from AI-Hypercomputer/bvandermoon-recipes
Update recipes to use new MaxText release and command
2 parents af2a7cd + 5092e25 commit f383c8f

File tree

16 files changed

+23
-23
lines changed

16 files changed

+23
-23
lines changed

‎training/trillium/GPT3-175B-MaxText/bf16/README.md‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -10,7 +10,7 @@ Please follow this [link](https://github.com/AI-Hypercomputer/tpu-recipes/blob/m
1010

1111
In step 1, use the MaxText [tpu-recipes-v0.1.0](https://github.com/AI-Hypercomputer/maxtext/releases/tag/tpu-recipes-v0.1.0) tag to run this recipe:
1212
```
13-
git checkout tpu-recipes-v0.1.0
13+
git checkout tpu-recipes-v0.1.1
1414
```
1515

1616
In step 2, use the jax-stable-stack image containing JAX 0.5.2:
@@ -25,7 +25,7 @@ bash docker_build_dependency_image.sh DEVICE=tpu MODE=stable_stack BASEIMAGE=${B
2525

2626
From the MaxText root directory, start your GPT3-175B workload
2727
```
28-
python3 benchmarks/benchmark_runner.py xpk \
28+
python3 -m benchmarks.benchmark_runner xpk \
2929
--project=$PROJECT \
3030
--zone=$ZONE \
3131
--device_type=v6e-256 \

‎training/trillium/GPT3-175B-MaxText/bf16/gpt3-175b-v6e-256.sh‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1,5 +1,5 @@
11
# Run this command from the MaxText root directory using the setup described in the README.
2-
python3 benchmarks/benchmark_runner.py xpk \
2+
python3 -m benchmarks.benchmark_runner xpk \
33
--project=$PROJECT \
44
--zone=$ZONE \
55
--device_type=v6e-256 \

‎training/trillium/Llama2-70B-MaxText/README.md‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -10,7 +10,7 @@ Please follow this [link](https://github.com/AI-Hypercomputer/tpu-recipes/blob/m
1010

1111
In step 1, use the MaxText [tpu-recipes-v0.1.0](https://github.com/AI-Hypercomputer/maxtext/releases/tag/tpu-recipes-v0.1.0) tag to run this recipe:
1212
```
13-
git checkout tpu-recipes-v0.1.0
13+
git checkout tpu-recipes-v0.1.1
1414
```
1515

1616
In step 2, use the jax-stable-stack image containing JAX 0.5.2:
@@ -25,7 +25,7 @@ bash docker_build_dependency_image.sh DEVICE=tpu MODE=stable_stack BASEIMAGE=${B
2525

2626
From the MaxText root directory, start your Llama2-70B workload
2727
```
28-
python3 benchmarks/benchmark_runner.py xpk \
28+
python3 -m benchmarks.benchmark_runner xpk \
2929
--project=$PROJECT \
3030
--zone=$ZONE \
3131
--device_type=v6e-256 \

‎training/trillium/Llama2-70B-MaxText/llama2-70b-v6e-256.sh‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1,5 +1,5 @@
11
# Run this command from the MaxText root directory using the setup described in the README.
2-
python3 benchmarks/benchmark_runner.py xpk \
2+
python3 -m benchmarks.benchmark_runner xpk \
33
--project=$PROJECT \
44
--zone=$ZONE \
55
--device_type=v6e-256 \

‎training/trillium/Llama3-8B-MaxText/v6e-8/README.md‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -10,7 +10,7 @@ Please follow this [link](https://github.com/AI-Hypercomputer/tpu-recipes/blob/m
1010

1111
In step 1, use the MaxText [tpu-recipes-v0.1.0](https://github.com/AI-Hypercomputer/maxtext/releases/tag/tpu-recipes-v0.1.0) tag to run this recipe:
1212
```
13-
git checkout tpu-recipes-v0.1.0
13+
git checkout tpu-recipes-v0.1.1
1414
```
1515

1616
In step 2, use the jax-stable-stack image containing JAX 0.5.2:
@@ -25,7 +25,7 @@ bash docker_build_dependency_image.sh DEVICE=tpu MODE=stable_stack BASEIMAGE=${B
2525

2626
From the MaxText root directory, start your Llama3.1-8B workload. Note: this benchmark uses a different model name than the equivalent v6e-256 recipe.
2727
```
28-
python3 benchmarks/benchmark_runner.py xpk \
28+
python3 -m benchmarks.benchmark_runner xpk \
2929
--project=$PROJECT \
3030
--zone=$ZONE \
3131
--device_type=v6e-8 \

‎training/trillium/Llama3-8B-MaxText/v6e-8/llama3-8B-1xv6e-8.sh‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1,5 +1,5 @@
11
# Run this command from the MaxText root directory using the setup described in the README.
2-
python3 benchmarks/benchmark_runner.py xpk \
2+
python3 -m benchmarks.benchmark_runner xpk \
33
--project=$PROJECT \
44
--zone=$ZONE \
55
--device_type=v6e-8 \

‎training/trillium/Llama3.1-405B-MaxText/README.md‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -10,7 +10,7 @@ Please follow this [link](https://github.com/AI-Hypercomputer/tpu-recipes/blob/m
1010

1111
In step 1, use the MaxText [tpu-recipes-v0.1.0](https://github.com/AI-Hypercomputer/maxtext/releases/tag/tpu-recipes-v0.1.0) tag to run this recipe:
1212
```
13-
git checkout tpu-recipes-v0.1.0
13+
git checkout tpu-recipes-v0.1.1
1414
```
1515

1616
In step 2, use the jax-stable-stack image containing JAX 0.5.2:
@@ -25,7 +25,7 @@ bash docker_build_dependency_image.sh DEVICE=tpu MODE=stable_stack BASEIMAGE=${B
2525

2626
From the MaxText root directory, start your Llama3.1-405B workload.
2727
```
28-
python3 benchmarks/benchmark_runner.py xpk \
28+
python3 -m benchmarks.benchmark_runner xpk \
2929
--project=$PROJECT \
3030
--zone=$ZONE \
3131
--device_type=v6e-256 \

‎training/trillium/Llama3.1-405B-MaxText/llama3-1-405b-2xv6e-256.sh‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1,5 +1,5 @@
11
# Run this command from the MaxText root directory using the setup described in the README.
2-
python3 benchmarks/benchmark_runner.py xpk \
2+
python3 -m benchmarks.benchmark_runner xpk \
33
--project=$PROJECT \
44
--zone=$ZONE \
55
--device_type=v6e-256 \

‎training/trillium/Llama3.1-70B-MaxText/README.md‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -10,7 +10,7 @@ Please follow this [link](https://github.com/AI-Hypercomputer/tpu-recipes/blob/m
1010

1111
In step 1, use the MaxText [tpu-recipes-v0.1.0](https://github.com/AI-Hypercomputer/maxtext/releases/tag/tpu-recipes-v0.1.0) tag to run this recipe:
1212
```
13-
git checkout tpu-recipes-v0.1.0
13+
git checkout tpu-recipes-v0.1.1
1414
```
1515

1616
In step 2, use the jax-stable-stack image containing JAX 0.5.2:
@@ -25,7 +25,7 @@ bash docker_build_dependency_image.sh DEVICE=tpu MODE=stable_stack BASEIMAGE=${B
2525

2626
From the MaxText root directory, start your Llama3.1-70B workload
2727
```
28-
python3 benchmarks/benchmark_runner.py xpk \
28+
python3 -m benchmarks.benchmark_runner xpk \
2929
--project=$PROJECT \
3030
--zone=$ZONE \
3131
--device_type=v6e-256 \

‎training/trillium/Llama3.1-70B-MaxText/llama3-1-70B-1xv6e-256.sh‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1,2 +1,2 @@
1-
python3 benchmarks/benchmark_runner.py --project=$PROJECT --zone=$ZONE --device_type=v6e-256 --num_slices=1 --cluster_name=${CLUSTER_NAME} --base_output_directory=${OUTPUT_DIR} \
1+
python3 -m benchmarks.benchmark_runner xpk --project=$PROJECT --zone=$ZONE --device_type=v6e-256 --num_slices=1 --cluster_name=${CLUSTER_NAME} --base_output_directory=${OUTPUT_DIR} \
22
--model_name="llama3_1_70b_8192" --base_docker_image maxtext_base_image

0 commit comments

Comments
 (0)