Merge branch 'stanford-crfm:main' into main

Author: NoushNabi

src/helm/config/model_deployments.yaml

@@ -1784,6 +1784,14 @@ model_deployments:
     client_spec:
       class_name: "helm.clients.together_client.TogetherClient"
 
+  - name: huggingface/olmo-1.7-7b
+    model_name: allenai/olmo-1.7-7b
+    tokenizer_name: allenai/OLMo-1.7-7B-hf
+    max_sequence_length: 2048
+    client_spec:
+      class_name: "helm.clients.huggingface_client.HuggingFaceClient"
+      args:
+        pretrained_model_name_or_path: allenai/OLMo-1.7-7B-hf
 
   ## MistralAI
   - name: together/mistral-7b-v0.1

src/helm/config/model_metadata.yaml

@@ -1460,6 +1460,14 @@ models:
     # TODO: Add instruct tag.
     tags: [TEXT_MODEL_TAG, FULL_FUNCTIONALITY_TEXT_MODEL_TAG]
 
+  - name: allenai/olmo-1.7-7b
+    display_name: OLMo 1.7 (7B)
+    description: OLMo is a series of Open Language Models trained on the Dolma dataset. The instruct versions was trained on the Tulu SFT mixture and a cleaned version of the UltraFeedback dataset.
+    creator_organization_name: Allen Institute for AI
+    access: open
+    num_parameters: 7000000000
+    release_date: 2024-04-17
+    tags: [TEXT_MODEL_TAG, FULL_FUNCTIONALITY_TEXT_MODEL_TAG]
 
   # Mistral AI
   - name: mistralai/mistral-7b-v0.1

src/helm/config/tokenizer_configs.yaml

@@ -310,6 +310,12 @@ tokenizer_configs:
     end_of_text_token: "<|endoftext|>"
     prefix_token: ""
 
+  - name: allenai/OLMo-1.7-7B-hf
+    tokenizer_spec:
+      class_name: "helm.tokenizers.huggingface_tokenizer.HuggingFaceTokenizer"
+    end_of_text_token: "<|endoftext|>"
+    prefix_token: ""
+
 
   # Microsoft
   - name: microsoft/phi-2