HuggingFaceH4-starchat-beta

assets/models/system/starchat-beta/asset.yaml

@@ -0,0 +1,4 @@
+extra_config: model.yaml
+spec: spec.yaml
+type: model
+categories: ["Foundation Models", "Text Generation"]

assets/models/system/starchat-beta/description.md

@@ -0,0 +1,62 @@
+# **Model Details**
+
+Name: HuggingFaceH4/starchat-beta
+StarChat is a series of language models designed as helpful coding assistants. StarChat-β, the second model in this series, is a fine-tuned version of StarCoderPlus. The fine-tuning utilizes an "uncensored" variant of the openassistant-guanaco dataset, enhancing performance on coding tasks. Note that this model may produce problematic outputs when prompted.
+
+Model Details:
+
+Model type: A 16B parameter GPT-like model fine-tuned on an "uncensored" variant of the openassistant-guanaco dataset.
+Language(s): Primarily English and supports 80+ programming languages.
+License: BigCode Open RAIL-M v1
+Finetuned from: bigcode/starcoderplus
+Model Sources:
+
+Intended Uses & Limitations:
+Intended for chat and coding assistance.
+Caution advised for potential problematic outputs.
+Bias and demographic skewing from GitHub community demographics.
+May produce syntactically valid but semantically incorrect code.
+Potential for generating false URLs that require careful inspection.
+Training and Evaluation:
+
+Trained on "uncensored" openassistant-guanaco dataset.
+Utilizes the ShareGPT filtering recipe behind the WizardLM.
+Hyperparameters, including learning rate, batch size, and others, specified for training.
+In summary, StarChat-β serves as a coding assistant, trained and fine-tuned for code-related tasks, offering capabilities in multiple programming languages. Users are cautioned about potential problematic outputs and advised to use the model responsibly for educational and research purposes.
+
+Repository: https://github.com/bigcode-project/starcoder
+Demo: https://huggingface.co/spaces/HuggingFaceH4/starchat-playground
+
+# **Inference samples**
+
+Inference type|Python sample (Notebook)|CLI with YAML
+|--|--|--|
+Real time|<a href="https://aka.ms/azureml-infer-online-sdk-text-generation-dolly" target="_blank">text-generation-online-endpoint.ipynb</a>|<a href="https://aka.ms/azureml-infer-online-cli-text-generation-dolly" target="_blank">text-generation-online-endpoint.sh</a>
+Batch |<a href="https://aka.ms/azureml-infer-batch-sdk-text-generation" target="_blank">text-generation-batch-endpoint.ipynb</a>| coming soon
+
+
+# **Sample inputs and outputs**
+
+### **Sample input**
+```json
+{
+    "input_data":{
+       "input_string":["the meaning of life is"],
+       "parameters":{
+             "temperature":0.5,
+             "top_p":0.5,
+             "max_new_tokens":10,
+              "do_sample":true
+       }
+    }
+}
+```
+
+### **Sample output**
+```json
+[
+  {
+    "0": "the meaning of life is to find your gift. the purpose of life is"
+  }
+]
+```

assets/models/system/starchat-beta/model.yaml

@@ -0,0 +1,8 @@
+path:
+  container_name: models
+  container_path: huggingface/HuggingFaceH4-starchat-beta/1706255740/mlflow_model_folder
+  storage_name: automlcesdkdataresources
+  type: azureblob
+publish:
+  description: description.md
+  type: mlflow_model

assets/models/system/starchat-beta/spec.yaml

@@ -0,0 +1,68 @@
+$schema: https://azuremlschemas.azureedge.net/latest/model.schema.json
+name: HuggingFaceH4-starchat-beta
+path: ./
+properties:
+  SharedComputeCapacityEnabled: true
+  SHA: b1bcda690655777373f57ea6614eb095ec2c886f
+  inference-min-sku-spec: 12|1|220|64
+  inference-recommended-sku: Standard_NC12s_v3, Standard_NC24s_v3, Standard_ND40rs_v2, Standard_NC24ads_A100_v4, Standard_NC48ads_A100_v4, Standard_NC96ads_A100_v4, Standard_ND96amsr_A100_v4, Standard_ND96asr_v4
+  evaluation-min-sku-spec: 6|1|112|128
+  evaluation-recommended-sku: Standard_NC6s_v3, Standard_NC12s_v3, Standard_NC24s_v3, Standard_NC24rs_v3, Standard_ND40rs_v2, Standard_NC48ads_A100_v4, Standard_NC96ads_A100_v4, Standard_ND96amsr_A100_v4, Standard_ND96asr_v4
+  finetune-min-sku-spec: 40|8|672|2900
+  finetune-recommended-sku: Standard_ND40rs_v2, Standard_ND96amsr_A100_v4, Standard_ND96asr_v4
+  finetuning-tasks: text-generation
+  languages : en
+  license : bigcode-openrail-m
+
+tags:
+  Featured: ""
+  SharedComputeCapacityEnabled: ""
+  huggingface_model_id: HuggingFaceH4/starchat-beta
+  inference_compute_allow_list:
+    [ 
+      STANDARD_DS4_V2
+      Standard_DS5_v2,
+      Standard_D16a_v4,
+      Standard_D16as_v4,
+      Standard_D32a_v4,
+      Standard_D32as_v4,
+      Standard_D48a_v4,
+      Standard_D48as_v4,
+      Standard_D64a_v4,
+      Standard_D64as_v4,
+      Standard_D96a_v4,
+      Standard_D96as_v4,
+      Standard_F4s_v2,
+      Standard_FX4mds,
+      Standard_F8s_v2,
+      Standard_FX12mds,
+      Standard_F16s_v2,
+      Standard_F32s_v2,
+      Standard_F48s_v2,
+      Standard_F64s_v2,
+      Standard_F72s_v2,
+      Standard_FX24mds,
+      Standard_FX36mds,
+      Standard_FX48mds,
+      Standard_E2s_v3,
+      Standard_E4s_v3,
+      Standard_E8s_v3,
+      Standard_E16s_v3,
+      Standard_E32s_v3,
+      Standard_E48s_v3,
+      Standard_E64s_v3,
+      Standard_NC4as_T4_v3,
+      Standard_NC6s_v3,
+      Standard_NC8as_T4_v3,
+      Standard_NC12s_v3,
+      Standard_NC16as_T4_v3,
+      Standard_NC24s_v3,
+      Standard_NC64as_T4_v3,
+      Standard_NC24ads_A100_v4,
+      Standard_NC48ads_A100_v4,
+      Standard_NC96ads_A100_v4,
+      Standard_ND96asr_v4,
+      Standard_ND96amsr_A100_v4,
+      Standard_ND40rs_v2,
+    ]    
+version: 1