Skip to content

update a4x recipe README #74

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

Open
ngu3 wants to merge 1 commit into
base: main
Choose a base branch
from
Open

update a4x recipe README #74

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
26 changes: 13 additions & 13 deletions ...ng/a4x/llama3-1-405b/nemo-pretraining-gke/16node-FP8CS-GBS2048/recipe/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -112,19 +112,19 @@ your client:
- To set the number of training steps to 100, run the following command from
your client:

```bash
cd $RECIPE_ROOT
export WORKLOAD_NAME=$USER-a4x-llama3-1-405b
helm install $WORKLOAD_NAME . -f values.yaml \
--set-file workload_launcher=launcher.sh \
--set-file workload_config=llama3-1-405b-fp8cs-gbs2048-gpus64.py \
--set workload.image=nvcr.io/nvidia/nemo:25.07 \
--set volumes.gcsMounts[0].bucketName=${GCS_BUCKET} \
--set volumes.gcsMounts[0].mountPath=/job-logs \
--set workload.envs[0].value=/job-logs/$WORKLOAD_NAME \
--set queue=${KUEUE_NAME} \
--set workload.arguments[0]="trainer.max_steps=100"
```
```bash
cd $RECIPE_ROOT
export WORKLOAD_NAME=$USER-a4x-llama3-1-405b
helm install $WORKLOAD_NAME . -f values.yaml \
--set-file workload_launcher=launcher.sh \
--set-file workload_config=llama3-1-405b-fp8cs-gbs2048-gpus64.py \
--set workload.image=nvcr.io/nvidia/nemo:25.07 \
--set volumes.gcsMounts[0].bucketName=${GCS_BUCKET} \
--set volumes.gcsMounts[0].mountPath=/job-logs \
--set workload.envs[0].value=/job-logs/$WORKLOAD_NAME \
--set queue=${KUEUE_NAME} \
--set workload.arguments[0]="trainer.max_steps=100"
```

### Monitor the job

Expand Down
24 changes: 12 additions & 12 deletions ...ning/a4x/llama3-1-70b/nemo-pretraining-gke/16node-BF16-GBS2048/recipe/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -87,18 +87,18 @@ gcloud container clusters get-credentials $CLUSTER_NAME --region $CLUSTER_REGION
To execute the job with the default settings, run the following command from
your client:

```bash
cd $RECIPE_ROOT
export WORKLOAD_NAME=$USER-a4x-llama3-1-70b-16node
helm install $WORKLOAD_NAME . -f values.yaml \
--set-file workload_launcher=launcher.sh \
--set-file workload_config=llama3-1-70b-bf16-gbs2048-gpus64.py \
--set workload.image=nvcr.io/nvidia/nemo:25.07 \
--set volumes.gcsMounts[0].bucketName=${GCS_BUCKET} \
--set volumes.gcsMounts[0].mountPath=/job-logs \
--set workload.envs[0].value=/job-logs/$WORKLOAD_NAME \
--set queue=${KUEUE_NAME}
```
```bash
cd $RECIPE_ROOT
export WORKLOAD_NAME=$USER-a4x-llama3-1-70b-16node
helm install $WORKLOAD_NAME . -f values.yaml \
--set-file workload_launcher=launcher.sh \
--set-file workload_config=llama3-1-70b-bf16-gbs2048-gpus64.py \
--set workload.image=nvcr.io/nvidia/nemo:25.07 \
--set volumes.gcsMounts[0].bucketName=${GCS_BUCKET} \
--set volumes.gcsMounts[0].mountPath=/job-logs \
--set workload.envs[0].value=/job-logs/$WORKLOAD_NAME \
--set queue=${KUEUE_NAME}
```

**Examples**

Expand Down
24 changes: 12 additions & 12 deletions ...ing/a4x/llama3-1-70b/nemo-pretraining-gke/16node-FP8CS-GBS2048/recipe/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -87,18 +87,18 @@ gcloud container clusters get-credentials $CLUSTER_NAME --region $CLUSTER_REGION
To execute the job with the default settings, run the following command from
your client:

```bash
cd $RECIPE_ROOT
export WORKLOAD_NAME=a4x-llama3-1-70b-fp8cs-gbs2048-gpus64
helm install $WORKLOAD_NAME . -f values.yaml \
--set-file workload_launcher=launcher.sh \
--set-file workload_config=llama3-1-70b-fp8cs-gbs2048-gpus64.py \
--set workload.image=nvcr.io/nvidia/nemo:25.07 \
--set volumes.gcsMounts[0].bucketName=${GCS_BUCKET} \
--set volumes.gcsMounts[0].mountPath=/job-logs \
--set workload.envs[0].value=/job-logs/$WORKLOAD_NAME \
--set queue=${KUEUE_NAME}
```
```bash
cd $RECIPE_ROOT
export WORKLOAD_NAME=a4x-llama3-1-70b-fp8cs-gbs2048-gpus64
helm install $WORKLOAD_NAME . -f values.yaml \
--set-file workload_launcher=launcher.sh \
--set-file workload_config=llama3-1-70b-fp8cs-gbs2048-gpus64.py \
--set workload.image=nvcr.io/nvidia/nemo:25.07 \
--set volumes.gcsMounts[0].bucketName=${GCS_BUCKET} \
--set volumes.gcsMounts[0].mountPath=/job-logs \
--set workload.envs[0].value=/job-logs/$WORKLOAD_NAME \
--set queue=${KUEUE_NAME}
```

**Examples**

Expand Down
24 changes: 12 additions & 12 deletions ...ing/a4x/llama3-1-70b/nemo-pretraining-gke/32node-FP8CS-GBS2048/recipe/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -87,18 +87,18 @@ gcloud container clusters get-credentials $CLUSTER_NAME --region $CLUSTER_REGION
To execute the job with the default settings, run the following command from
your client:

```bash
cd $RECIPE_ROOT
export WORKLOAD_NAME=$USER-a4x-llama3-1-70b-32node
helm install $WORKLOAD_NAME . -f values.yaml \
--set-file workload_launcher=launcher.sh \
--set-file workload_config=llama3-1-70b-fp8cs-gbs2048-gpus128.py \
--set workload.image=nvcr.io/nvidia/nemo:25.07 \
--set volumes.gcsMounts[0].bucketName=${GCS_BUCKET} \
--set volumes.gcsMounts[0].mountPath=/job-logs \
--set workload.envs[0].value=/job-logs/$WORKLOAD_NAME \
--set queue=${KUEUE_NAME}
```
```bash
cd $RECIPE_ROOT
export WORKLOAD_NAME=$USER-a4x-llama3-1-70b-32node
helm install $WORKLOAD_NAME . -f values.yaml \
--set-file workload_launcher=launcher.sh \
--set-file workload_config=llama3-1-70b-fp8cs-gbs2048-gpus128.py \
--set workload.image=nvcr.io/nvidia/nemo:25.07 \
--set volumes.gcsMounts[0].bucketName=${GCS_BUCKET} \
--set volumes.gcsMounts[0].mountPath=/job-logs \
--set workload.envs[0].value=/job-logs/$WORKLOAD_NAME \
--set queue=${KUEUE_NAME}
```

**Examples**

Expand Down
24 changes: 12 additions & 12 deletions ...ing/a4x/llama3-1-70b/nemo-pretraining-gke/64node-FP8CS-GBS2048/recipe/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -87,18 +87,18 @@ gcloud container clusters get-credentials $CLUSTER_NAME --region $CLUSTER_REGION
To execute the job with the default settings, run the following command from
your client:

```bash
cd $RECIPE_ROOT
export WORKLOAD_NAME=$USER-a4x-llama3-1-70b-64node
helm install $WORKLOAD_NAME . -f values.yaml \
--set-file workload_launcher=launcher.sh \
--set-file workload_config=llama3-1-70b-fp8cs-gbs2048-gpus256.py \
--set workload.image=nvcr.io/nvidia/nemo:25.07 \
--set volumes.gcsMounts[0].bucketName=${GCS_BUCKET} \
--set volumes.gcsMounts[0].mountPath=/job-logs \
--set workload.envs[0].value=/job-logs/$WORKLOAD_NAME \
--set queue=${KUEUE_NAME}
```
```bash
cd $RECIPE_ROOT
export WORKLOAD_NAME=$USER-a4x-llama3-1-70b-64node
helm install $WORKLOAD_NAME . -f values.yaml \
--set-file workload_launcher=launcher.sh \
--set-file workload_config=llama3-1-70b-fp8cs-gbs2048-gpus256.py \
--set workload.image=nvcr.io/nvidia/nemo:25.07 \
--set volumes.gcsMounts[0].bucketName=${GCS_BUCKET} \
--set volumes.gcsMounts[0].mountPath=/job-logs \
--set workload.envs[0].value=/job-logs/$WORKLOAD_NAME \
--set queue=${KUEUE_NAME}
```

**Examples**

Expand Down
24 changes: 12 additions & 12 deletions training/a4x/llama3-1-8b/nemo-pretraining-gke/16node-BF16-GBS1024/recipe/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -87,18 +87,18 @@ gcloud container clusters get-credentials $CLUSTER_NAME --region $CLUSTER_REGION
To execute the job with the default settings, run the following command from
your client:

```bash
cd $RECIPE_ROOT
export WORKLOAD_NAME=$USER-a4x-llama3-1-8b-16node
helm install $WORKLOAD_NAME . -f values.yaml \
--set-file workload_launcher=launcher.sh \
--set-file workload_config=llama3-1-8b-bf16-gbs1024-gpus64.py \
--set workload.image=nvcr.io/nvidia/nemo:25.07 \
--set volumes.gcsMounts[0].bucketName=${GCS_BUCKET} \
--set volumes.gcsMounts[0].mountPath=/job-logs \
--set workload.envs[0].value=/job-logs/$WORKLOAD_NAME \
--set queue=${KUEUE_NAME}
```
```bash
cd $RECIPE_ROOT
export WORKLOAD_NAME=$USER-a4x-llama3-1-8b-16node
helm install $WORKLOAD_NAME . -f values.yaml \
--set-file workload_launcher=launcher.sh \
--set-file workload_config=llama3-1-8b-bf16-gbs1024-gpus64.py \
--set workload.image=nvcr.io/nvidia/nemo:25.07 \
--set volumes.gcsMounts[0].bucketName=${GCS_BUCKET} \
--set volumes.gcsMounts[0].mountPath=/job-logs \
--set workload.envs[0].value=/job-logs/$WORKLOAD_NAME \
--set queue=${KUEUE_NAME}
```

**Examples**

Expand Down
24 changes: 12 additions & 12 deletions training/a4x/llama3-1-8b/nemo-pretraining-gke/16node-FP8CS-GBS128/recipe/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -86,18 +86,18 @@ gcloud container clusters get-credentials $CLUSTER_NAME --region $CLUSTER_REGION
To execute the job with the default settings, run the following command from
your client:

```bash
cd $RECIPE_ROOT
export WORKLOAD_NAME=a4x-llama3-1-8b-fp8cs-gbs128-gpus64-16node
helm install $WORKLOAD_NAME . -f values.yaml \
--set-file workload_launcher=launcher.sh \
--set-file workload_config=llama3-1-8b-fp8cs-gbs128-gpus64.py \
--set workload.image=nvcr.io/nvidia/nemo:25.07 \
--set volumes.gcsMounts[0].bucketName=${GCS_BUCKET} \
--set volumes.gcsMounts[0].mountPath=/job-logs \
--set workload.envs[0].value=/job-logs/$WORKLOAD_NAME \
--set queue=${KUEUE_NAME}
```
```bash
cd $RECIPE_ROOT
export WORKLOAD_NAME=a4x-llama3-1-8b-fp8cs-gbs128-gpus64-16node
helm install $WORKLOAD_NAME . -f values.yaml \
--set-file workload_launcher=launcher.sh \
--set-file workload_config=llama3-1-8b-fp8cs-gbs128-gpus64.py \
--set workload.image=nvcr.io/nvidia/nemo:25.07 \
--set volumes.gcsMounts[0].bucketName=${GCS_BUCKET} \
--set volumes.gcsMounts[0].mountPath=/job-logs \
--set workload.envs[0].value=/job-logs/$WORKLOAD_NAME \
--set queue=${KUEUE_NAME}
```

**Examples**

Expand Down
24 changes: 12 additions & 12 deletions ...ng/a4x/nemotron4-340B/nemo-pretraining-gke/32node-FP8CS-GBS256/recipe/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -87,18 +87,18 @@ gcloud container clusters get-credentials $CLUSTER_NAME --region $CLUSTER_REGION
To execute the job with the default settings, run the following command from
your client:

```bash
cd $RECIPE_ROOT
export WORKLOAD_NAME=$USER-a4x-nemotron4-340b-32node
helm install $WORKLOAD_NAME . -f values.yaml \
--set-file workload_launcher=launcher.sh \
--set-file workload_config=nemotron4-340b-fp8cs-gbs256-gpus128.py \
--set workload.image=nvcr.io/nvidia/nemo:25.09 \
--set volumes.gcsMounts[0].bucketName=${GCS_BUCKET} \
--set volumes.gcsMounts[0].mountPath=/job-logs \
--set workload.envs[0].value=/job-logs/$WORKLOAD_NAME \
--set queue=${KUEUE_NAME}
```
```bash
cd $RECIPE_ROOT
export WORKLOAD_NAME=$USER-a4x-nemotron4-340b-32node
helm install $WORKLOAD_NAME . -f values.yaml \
--set-file workload_launcher=launcher.sh \
--set-file workload_config=nemotron4-340b-fp8cs-gbs256-gpus128.py \
--set workload.image=nvcr.io/nvidia/nemo:25.09 \
--set volumes.gcsMounts[0].bucketName=${GCS_BUCKET} \
--set volumes.gcsMounts[0].mountPath=/job-logs \
--set workload.envs[0].value=/job-logs/$WORKLOAD_NAME \
--set queue=${KUEUE_NAME}
```

**Examples**

Expand Down