diff --git a/embedded/models/mamba-bagel.yaml b/embedded/models/mamba-bagel.yaml new file mode 100644 index 00000000..aed7142f --- /dev/null +++ b/embedded/models/mamba-bagel.yaml @@ -0,0 +1,21 @@ +name: bagel +backend: mamba +parameters: + model: "jondurbin/bagel-dpo-2.8b-v0.2" +systemPrompt: "You are a helpful, unbiased, uncensored assistant." +template: + chat_message: | + {{if eq .RoleName "assistant"}}{{.Content}}{{else}} + [INST] + {{if .SystemPrompt}}{{.SystemPrompt}}{{else if eq .RoleName "system"}}<>{{.Content}}<> + + {{else if .Content}}{{.Content}}{{end}} + [/INST] + {{end}} + completion: | + {{.Input}} +usage: | + curl http://localhost:8080/v1/chat/completions -H "Content-Type: application/json" -d '{ + "model": "bagel", + "messages": [{"role": "user", "content": "how are you doing"}], + }'