fix: miscelleanous updates

kaito-project · Jan 26, 2024 · e33e56d · e33e56d
1 parent cf325ea
commit e33e56d
Show file tree

Hide file tree

Showing 4 changed files with 12 additions and 6 deletions.
diff --git a/.github/workflows/e2e-preset-test.yml b/.github/workflows/e2e-preset-test.yml
@@ -108,9 +108,9 @@ jobs:
         run: |
             # Set the ACR based on the tag value
             if [[ "${{ needs.setup.outputs.image_tag }}" == "latest" ]]; then
-              echo "ACR_NAME=aimodelsregistry" >> $GITHUB_OUTPUT
+              echo "ACR_NAME=${{secrets.ACR_AMR_USERNAME}}" >> $GITHUB_OUTPUT
             else
-              echo "ACR_NAME=aimodelsregistrytest" >> $GITHUB_OUTPUT
+              echo "ACR_NAME=${{secrets.ACR_AMRT_USERNAME}}" >> $GITHUB_OUTPUT
             fi
         
       - name: Install Azure CLI latest

diff --git a/presets/models/falcon/model.go b/presets/models/falcon/model.go
@@ -36,8 +36,12 @@ var (
 	PresetFalcon7BInstructModel  = PresetFalcon7BModel + "-instruct"
 	PresetFalcon40BInstructModel = PresetFalcon40BModel + "-instruct"
 
-	baseCommandPresetFalcon = "accelerate launch --use_deepspeed"
-	falconRunParams         = map[string]string{}
+	baseCommandPresetFalcon = "accelerate launch"
+
+	falconRunParams = map[string]string{
+		"torch_dtype": "bfloat16",
+		"pipeline":    "text-generation",
+	}
 )
 
 var falconA falcon7b

diff --git a/presets/models/llama2/README.md b/presets/models/llama2/README.md
@@ -31,7 +31,8 @@ Use the following command to build the llama2 inference service image from the r
 docker build \
   --file docker/presets/llama-2/Dockerfile \
   --build-arg WEIGHTS_PATH=$LLAMA_WEIGHTS_PATH \
-  --build-arg MODEL_PRESET_PATH=presets/models/llama2 \
+  --build-arg MODEL_TYPE=llama2-completion \
+  --build-arg VERSION=0.0.1 \
   -t $LLAMA_MODEL_NAME:latest .
 ```
 

diff --git a/presets/models/llama2chat/README.md b/presets/models/llama2chat/README.md
@@ -31,7 +31,8 @@ Use the following command to build the llama2chat inference service image from t
 docker build \
   --file docker/presets/llama-2/Dockerfile \
   --build-arg WEIGHTS_PATH=$LLAMA_WEIGHTS_PATH \
-  --build-arg MODEL_PRESET_PATH=presets/models/llama2chat \
+  --build-arg MODEL_TYPE=llama2-chat \
+  --build-arg VERSION=0.0.1 \
   -t $LLAMA_MODEL_NAME:latest .
 ```