Update local metadata to include new directoryPath and promptTemplate tags (#4152)

sophies927 · web-flow · commit 52d8d22d8d9b · 2025-05-07T19:33:03.000-07:00
* Add directoryPath and promptTemplate for models that have them

* Add remaining directoryPath tags

* Add missing Phi model promptTemplate tags

* Add missing Phi-4 and DeepSeek promptTemplate tags

* Remove Phi-4-mini prompt templates for now

* Fix promptTemplates for Phi-4 models
diff --git a/assets/models/system/DeepSeek-R1-Distill-Qwen-1.5B-cuda-gpu/spec.yaml b/assets/models/system/DeepSeek-R1-Distill-Qwen-1.5B-cuda-gpu/spec.yaml
@@ -12,6 +12,8 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: deepseek-r1-1.5b
+  directoryPath: cuda-int4-rtn-block-32
+  promptTemplate: "{\"assistant\": \"{Content}\", \"prompt\": \"\\\\u003C\\\\uFF5CUser\\\\uFF5C\\\\u003E{Content}\\\\u003C\\\\uFF5CAssistant\\\\uFF5C\\\\u003E\"}"
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/DeepSeek-R1-Distill-Qwen-1.5B-generic-cpu/spec.yaml b/assets/models/system/DeepSeek-R1-Distill-Qwen-1.5B-generic-cpu/spec.yaml
@@ -12,6 +12,8 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: deepseek-r1-1.5b
+  directoryPath: cpu-int4-rtn-block-32-acc-level-4
+  promptTemplate: "{\"assistant\": \"{Content}\", \"prompt\": \"\\\\u003C\\\\uFF5CUser\\\\uFF5C\\\\u003E{Content}\\\\u003C\\\\uFF5CAssistant\\\\uFF5C\\\\u003E\"}"
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/DeepSeek-R1-Distill-Qwen-1.5B-generic-gpu/spec.yaml b/assets/models/system/DeepSeek-R1-Distill-Qwen-1.5B-generic-gpu/spec.yaml
@@ -12,6 +12,8 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: deepseek-r1-1.5b
+  directoryPath: directml-int4-rtn-block-32-acc-level-4
+  promptTemplate: "{\"assistant\": \"{Content}\", \"prompt\": \"\\\\u003C\\\\uFF5CUser\\\\uFF5C\\\\u003E{Content}\\\\u003C\\\\uFF5CAssistant\\\\uFF5C\\\\u003E\"}"
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/DeepSeek-R1-Distill-Qwen-7B-cuda-gpu/spec.yaml b/assets/models/system/DeepSeek-R1-Distill-Qwen-7B-cuda-gpu/spec.yaml
@@ -12,6 +12,8 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: deepseek-r1-7b
+  directoryPath: cuda-int4-rtn-block-32
+  promptTemplate: "{\"assistant\": \"{Content}\", \"prompt\": \"\\\\u003C\\\\uFF5CUser\\\\uFF5C\\\\u003E{Content}\\\\u003C\\\\uFF5CAssistant\\\\uFF5C\\\\u003E\"}"
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/DeepSeek-R1-Distill-Qwen-7B-generic-cpu/spec.yaml b/assets/models/system/DeepSeek-R1-Distill-Qwen-7B-generic-cpu/spec.yaml
@@ -12,6 +12,8 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: deepseek-r1-7b
+  directoryPath: cpu-int4-rtn-block-32-acc-level-4
+  promptTemplate: "{\"assistant\": \"{Content}\", \"prompt\": \"\\\\u003C\\\\uFF5CUser\\\\uFF5C\\\\u003E{Content}\\\\u003C\\\\uFF5CAssistant\\\\uFF5C\\\\u003E\"}"
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/DeepSeek-R1-Distill-Qwen-7B-generic-gpu/spec.yaml b/assets/models/system/DeepSeek-R1-Distill-Qwen-7B-generic-gpu/spec.yaml
@@ -12,6 +12,8 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: deepseek-r1-7b
+  directoryPath: directml-int4-rtn-block-32-acc-level-4
+  promptTemplate: "{\"assistant\": \"{Content}\", \"prompt\": \"\\\\u003C\\\\uFF5CUser\\\\uFF5C\\\\u003E{Content}\\\\u003C\\\\uFF5CAssistant\\\\uFF5C\\\\u003E\"}"
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/Mistral-7B-Instruct-v0-2-cuda-gpu/spec.yaml b/assets/models/system/Mistral-7B-Instruct-v0-2-cuda-gpu/spec.yaml
@@ -12,6 +12,8 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: mistral-7b-v0.2
+  directoryPath: mistral-7b-instruct-v0.2-cuda-int4-rtn-block-32
+  promptTemplate: "{\"system\": \"<s>\", \"user\": \"[INST]\\n{Content}\\n[/INST]\", \"assistant\": \"{Content}</s>\", \"prompt\": \"[INST]\\n{Content}\\n[/INST]\"}"
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/Mistral-7B-Instruct-v0-2-generic-cpu/spec.yaml b/assets/models/system/Mistral-7B-Instruct-v0-2-generic-cpu/spec.yaml
@@ -12,6 +12,8 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: mistral-7b-v0.2
+  directoryPath: mistral-7b-instruct-v0.2-cpu-int4-rtn-block-32
+  promptTemplate: "{\"system\": \"<s>\", \"user\": \"[INST]\\n{Content}\\n[/INST]\", \"assistant\": \"{Content}</s>\", \"prompt\": \"[INST]\\n{Content}\\n[/INST]\"}"
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/Mistral-7B-Instruct-v0-2-generic-gpu/spec.yaml b/assets/models/system/Mistral-7B-Instruct-v0-2-generic-gpu/spec.yaml
@@ -12,6 +12,8 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: mistral-7b-v0.2
+  directoryPath: mistralai_Mistral-7B-Instruct-v0.2
+  promptTemplate: "{\"system\": \"<s>\", \"user\": \"[INST]\\n{Content}\\n[/INST]\", \"assistant\": \"{Content}</s>\", \"prompt\": \"[INST]\\n{Content}\\n[/INST]\"}"
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/Phi-4-cuda-gpu/spec.yaml b/assets/models/system/Phi-4-cuda-gpu/spec.yaml
@@ -12,6 +12,7 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: phi-4
+  directoryPath: cuda-int4-rtn-block-32
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/Phi-4-generic-cpu/spec.yaml b/assets/models/system/Phi-4-generic-cpu/spec.yaml
@@ -12,6 +12,7 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: phi-4
+  directoryPath: cpu-int4-rtn-block-32-acc-level-4
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/Phi-4-generic-gpu/spec.yaml b/assets/models/system/Phi-4-generic-gpu/spec.yaml
@@ -12,6 +12,7 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: phi-4
+  directoryPath: directml-int4-rtn-block-32-acc-level-4
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/deepseek-r1-distill-llama-8b-cuda-gpu/spec.yaml b/assets/models/system/deepseek-r1-distill-llama-8b-cuda-gpu/spec.yaml
@@ -12,6 +12,8 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: deepseek-r1-8b
+  directoryPath: cuda-int4-rtn-block-32
+  promptTemplate: "{\"assistant\": \"{Content}\", \"prompt\": \"\\\\u003C\\\\uFF5CUser\\\\uFF5C\\\\u003E{Content}\\\\u003C\\\\uFF5CAssistant\\\\uFF5C\\\\u003E\"}"
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/deepseek-r1-distill-llama-8b-generic-cpu/spec.yaml b/assets/models/system/deepseek-r1-distill-llama-8b-generic-cpu/spec.yaml
@@ -12,6 +12,8 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: deepseek-r1-8b
+  directoryPath: cpu-int4-rtn-block-32-acc-level-4
+  promptTemplate: "{\"assistant\": \"{Content}\", \"prompt\": \"\\\\u003C\\\\uFF5CUser\\\\uFF5C\\\\u003E{Content}\\\\u003C\\\\uFF5CAssistant\\\\uFF5C\\\\u003E\"}"
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/deepseek-r1-distill-llama-8b-generic-gpu/spec.yaml b/assets/models/system/deepseek-r1-distill-llama-8b-generic-gpu/spec.yaml
@@ -12,6 +12,8 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: deepseek-r1-8b
+  directoryPath: directml-int4-rtn-block-32-acc-level-4
+  promptTemplate: "{\"assistant\": \"{Content}\", \"prompt\": \"\\\\u003C\\\\uFF5CUser\\\\uFF5C\\\\u003E{Content}\\\\u003C\\\\uFF5CAssistant\\\\uFF5C\\\\u003E\"}"
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/deepseek-r1-distill-qwen-14b-cuda-gpu/spec.yaml b/assets/models/system/deepseek-r1-distill-qwen-14b-cuda-gpu/spec.yaml
@@ -12,6 +12,8 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: deepseek-r1-14b
+  directoryPath: cuda-int4-rtn-block-32
+  promptTemplate: "{\"assistant\": \"{Content}\", \"prompt\": \"\\\\u003C\\\\uFF5CUser\\\\uFF5C\\\\u003E{Content}\\\\u003C\\\\uFF5CAssistant\\\\uFF5C\\\\u003E\"}"
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/deepseek-r1-distill-qwen-14b-generic-cpu/spec.yaml b/assets/models/system/deepseek-r1-distill-qwen-14b-generic-cpu/spec.yaml
@@ -12,6 +12,8 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: deepseek-r1-14b
+  directoryPath: cpu-int4-rtn-block-32-acc-level-4
+  promptTemplate: "{\"assistant\": \"{Content}\", \"prompt\": \"\\\\u003C\\\\uFF5CUser\\\\uFF5C\\\\u003E{Content}\\\\u003C\\\\uFF5CAssistant\\\\uFF5C\\\\u003E\"}"
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/deepseek-r1-distill-qwen-14b-generic-gpu/spec.yaml b/assets/models/system/deepseek-r1-distill-qwen-14b-generic-gpu/spec.yaml
@@ -12,6 +12,8 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: deepseek-r1-14b
+  directoryPath: directml-int4-rtn-block-32-acc-level-4
+  promptTemplate: "{\"assistant\": \"{Content}\", \"prompt\": \"\\\\u003C\\\\uFF5CUser\\\\uFF5C\\\\u003E{Content}\\\\u003C\\\\uFF5CAssistant\\\\uFF5C\\\\u003E\"}"
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/phi-3-mini-128k-instruct-cuda-gpu/spec.yaml b/assets/models/system/phi-3-mini-128k-instruct-cuda-gpu/spec.yaml
@@ -12,6 +12,8 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: phi-3-mini-128k
+  directoryPath: phi-3-mini-128k-instruct-cuda-int4-rtn-block-32
+  promptTemplate: "{\"system\": \"<|system|>\\n{Content}<|end|>\", \"user\": \"<|user|>\\n{Content}<|end|>\", \"assistant\": \"<|assistant|>\\n{Content}<|end|>\", \"prompt\": \"<|user|>\\n{Content}<|end|>\\n<|assistant|>\"}"
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/phi-3-mini-128k-instruct-generic-cpu/spec.yaml b/assets/models/system/phi-3-mini-128k-instruct-generic-cpu/spec.yaml
@@ -12,6 +12,8 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: phi-3-mini-128k
+  directoryPath: phi-3-mini-128k-instruct-cpu-int4-rtn-block-32
+  promptTemplate: "{\"system\": \"<|system|>\\n{Content}<|end|>\", \"user\": \"<|user|>\\n{Content}<|end|>\", \"assistant\": \"<|assistant|>\\n{Content}<|end|>\", \"prompt\": \"<|user|>\\n{Content}<|end|>\\n<|assistant|>\"}"
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/phi-3-mini-4k-instruct-cuda-gpu/spec.yaml b/assets/models/system/phi-3-mini-4k-instruct-cuda-gpu/spec.yaml
@@ -12,6 +12,8 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: phi-3-mini-4k
+  directoryPath: phi-3-mini-4k-instruct-cuda-int4-rtn-block-32
+  promptTemplate: "{\"system\": \"<|system|>\\n{Content}<|end|>\", \"user\": \"<|user|>\\n{Content}<|end|>\", \"assistant\": \"<|assistant|>\\n{Content}<|end|>\", \"prompt\": \"<|user|>\\n{Content}<|end|>\\n<|assistant|>\"}"
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/phi-3-mini-4k-instruct-generic-cpu/spec.yaml b/assets/models/system/phi-3-mini-4k-instruct-generic-cpu/spec.yaml
@@ -12,6 +12,8 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: phi-3-mini-4k
+  directoryPath: phi-3-mini-4k-instruct-cpu-int4-rtn-block-32
+  promptTemplate: "{\"system\": \"<|system|>\\n{Content}<|end|>\", \"user\": \"<|user|>\\n{Content}<|end|>\", \"assistant\": \"<|assistant|>\\n{Content}<|end|>\", \"prompt\": \"<|user|>\\n{Content}<|end|>\\n<|assistant|>\"}"
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/phi-3.5-mini-128k-instruct-cuda-gpu/spec.yaml b/assets/models/system/phi-3.5-mini-128k-instruct-cuda-gpu/spec.yaml
@@ -12,6 +12,8 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: phi-3.5-mini
+  directoryPath: cuda-int4-rtn-block-32
+  promptTemplate: "{\"system\": \"<|system|>\\n{Content}<|end|>\", \"user\": \"<|user|>\\n{Content}<|end|>\", \"assistant\": \"<|assistant|>\\n{Content}<|end|>\", \"prompt\": \"<|user|>\\n{Content}<|end|>\\n<|assistant|>\"}"
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/phi-3.5-mini-128k-instruct-generic-cpu/spec.yaml b/assets/models/system/phi-3.5-mini-128k-instruct-generic-cpu/spec.yaml
@@ -12,6 +12,8 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: phi-3.5-mini
+  directoryPath: cpu-int4-rtn-block-32-acc-level-4
+  promptTemplate: "{\"system\": \"<|system|>\\n{Content}<|end|>\", \"user\": \"<|user|>\\n{Content}<|end|>\", \"assistant\": \"<|assistant|>\\n{Content}<|end|>\", \"prompt\": \"<|user|>\\n{Content}<|end|>\\n<|assistant|>\"}"
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/phi-3.5-mini-128k-instruct-generic-gpu/spec.yaml b/assets/models/system/phi-3.5-mini-128k-instruct-generic-gpu/spec.yaml
@@ -12,6 +12,8 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: phi-3.5-mini
+  directoryPath: directml-int4-rtn-block-32-acc-level-4
+  promptTemplate: "{\"system\": \"<|system|>\\n{Content}<|end|>\", \"user\": \"<|user|>\\n{Content}<|end|>\", \"assistant\": \"<|assistant|>\\n{Content}<|end|>\", \"prompt\": \"<|user|>\\n{Content}<|end|>\\n<|assistant|>\"}"
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/phi-4-mini-instruct-cuda-gpu/spec.yaml b/assets/models/system/phi-4-mini-instruct-cuda-gpu/spec.yaml
@@ -12,6 +12,8 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: phi-4-mini
+  directoryPath: cuda-int4-int8
+  promptTemplate: "{\"system\": \"<|system|>\\n{Content}<|end|>\", \"user\": \"<|user|>\\n{Content}<|end|>\", \"assistant\": \"<|assistant|>\\n{Content}<|end|>\", \"prompt\": \"<|user|>\\n{Content}<|end|>\\n<|assistant|>\", \"tool_spec\": \"<|tool|>\\n{Content}<|end|>\", \"tool\": \"<|tool_response|>\\n{Content}<|end|>\", \"tool_prompt\": \"<|tool_response|>\\n{Content}<|end|>\\n<|assistant|>\"}"
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/phi-4-mini-instruct-generic-cpu/spec.yaml b/assets/models/system/phi-4-mini-instruct-generic-cpu/spec.yaml
@@ -12,6 +12,8 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: phi-4-mini
+  directoryPath: cpu-int4-int8
+  promptTemplate: "{\"system\": \"<|system|>\\n{Content}<|end|>\", \"user\": \"<|user|>\\n{Content}<|end|>\", \"assistant\": \"<|assistant|>\\n{Content}<|end|>\", \"prompt\": \"<|user|>\\n{Content}<|end|>\\n<|assistant|>\", \"tool_spec\": \"<|tool|>\\n{Content}<|end|>\", \"tool\": \"<|tool_response|>\\n{Content}<|end|>\", \"tool_prompt\": \"<|tool_response|>\\n{Content}<|end|>\\n<|assistant|>\"}"
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/qwen2.5-0.5b-instruct-cuda-gpu/spec.yaml b/assets/models/system/qwen2.5-0.5b-instruct-cuda-gpu/spec.yaml
@@ -12,6 +12,7 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: qwen2.5-0.5b
+  directoryPath: cuda-int4-rtn-block-32
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/qwen2.5-0.5b-instruct-generic-cpu/spec.yaml b/assets/models/system/qwen2.5-0.5b-instruct-generic-cpu/spec.yaml
@@ -12,6 +12,7 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: qwen2.5-0.5b
+  directoryPath: cpu-int4-rtn-block-32-acc-level-4
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/qwen2.5-0.5b-instruct-generic-gpu/spec.yaml b/assets/models/system/qwen2.5-0.5b-instruct-generic-gpu/spec.yaml
@@ -12,6 +12,7 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: qwen2.5-0.5b
+  directoryPath: directml-int4-awq-block-128
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/qwen2.5-1.5b-instruct-cuda-gpu/spec.yaml b/assets/models/system/qwen2.5-1.5b-instruct-cuda-gpu/spec.yaml
@@ -12,6 +12,7 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: qwen2.5-1.5b
+  directoryPath: cuda-int4-rtn-block-32
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/qwen2.5-1.5b-instruct-generic-cpu/spec.yaml b/assets/models/system/qwen2.5-1.5b-instruct-generic-cpu/spec.yaml
@@ -12,6 +12,7 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: qwen2.5-1.5b
+  directoryPath: cpu-int4-rtn-block-32-acc-level-4
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/qwen2.5-1.5b-instruct-generic-gpu/spec.yaml b/assets/models/system/qwen2.5-1.5b-instruct-generic-gpu/spec.yaml
@@ -12,6 +12,7 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: qwen2.5-1.5b
+  directoryPath: directml-int4-awq-block-128
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/qwen2.5-14b-instruct-cuda-gpu/spec.yaml b/assets/models/system/qwen2.5-14b-instruct-cuda-gpu/spec.yaml
@@ -12,6 +12,7 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: qwen2.5-14b
+  directoryPath: cuda-int4-rtn-block-32
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/qwen2.5-14b-instruct-generic-cpu/spec.yaml b/assets/models/system/qwen2.5-14b-instruct-generic-cpu/spec.yaml
@@ -12,6 +12,7 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: qwen2.5-14b
+  directoryPath: cpu-int4-rtn-block-32-acc-level-4
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/qwen2.5-14b-instruct-generic-gpu/spec.yaml b/assets/models/system/qwen2.5-14b-instruct-generic-gpu/spec.yaml
@@ -12,6 +12,7 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: qwen2.5-14b
+  directoryPath: directml-int4-awq-block-128
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/qwen2.5-3b-instruct-cuda-gpu/spec.yaml b/assets/models/system/qwen2.5-3b-instruct-cuda-gpu/spec.yaml
@@ -12,6 +12,7 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: qwen2.5-3b
+  directoryPath: cuda-int4-rtn-block-32
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/qwen2.5-3b-instruct-generic-cpu/spec.yaml b/assets/models/system/qwen2.5-3b-instruct-generic-cpu/spec.yaml
@@ -12,6 +12,7 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: qwen2.5-3b
+  directoryPath: cpu-int4-rtn-block-32-acc-level-4
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/qwen2.5-3b-instruct-generic-gpu/spec.yaml b/assets/models/system/qwen2.5-3b-instruct-generic-gpu/spec.yaml
@@ -12,6 +12,7 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: qwen2.5-3b
+  directoryPath: directml-int4-awq-block-128
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/qwen2.5-7b-instruct-cuda-gpu/spec.yaml b/assets/models/system/qwen2.5-7b-instruct-cuda-gpu/spec.yaml
@@ -12,6 +12,7 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: qwen2.5-7b
+  directoryPath: cuda-int4-rtn-block-32
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/qwen2.5-7b-instruct-generic-cpu/spec.yaml b/assets/models/system/qwen2.5-7b-instruct-generic-cpu/spec.yaml
@@ -12,6 +12,7 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: qwen2.5-7b
+  directoryPath: cpu-int4-rtn-block-32-acc-level-4
 type: custom_model
 variantInfo:
   parents:
diff --git a/assets/models/system/qwen2.5-7b-instruct-generic-gpu/spec.yaml b/assets/models/system/qwen2.5-7b-instruct-generic-gpu/spec.yaml
@@ -12,6 +12,7 @@ tags:
   task: chat-completion
   maxOutputTokens: 2048
   alias: qwen2.5-7b
+  directoryPath: directml-int4-awq-block-128
 type: custom_model
 variantInfo:
   parents:
