Add claude, gemini, and palm LLM charts and configuration (#61)

Update MQ deployments to support multiple mounted secrets for Google API compat. Add base personas for added LLMs to test Update backend deployment script to include new LLM services config Co-authored-by: Daniel McKnight <[email protected]>
NeonGeckoCom · Jan 17, 2024 · 62d3eee · 62d3eee
1 parent 78c1bdf
commit 62d3eee
Show file tree

Hide file tree

Showing 39 changed files with 554 additions and 31 deletions.
diff --git a/neon_diana_utils/configuration.py b/neon_diana_utils/configuration.py
@@ -87,7 +87,12 @@ def make_llm_bot_config():
     configuration = {"llm_bots": dict()}
     if click.confirm("Configure ChatGPT Personas?"):
         configuration['llm_bots']['chat_gpt'] = persona_config['chat_gpt']
-    # TODO: Add other LLM personas here
+    if click.confirm("Configure PaLM2 Personas?"):
+        configuration['llm_bots']['palm2'] = persona_config['palm2']
+    if click.confirm("Configure Gemini Personas?"):
+        configuration['llm_bots']['gemini'] = persona_config['gemini']
+    if click.confirm("Configure Claude Personas?"):
+        configuration['llm_bots']['claude'] = persona_config['claude']
     return configuration
 
 
@@ -163,7 +168,7 @@ def make_keys_config(write_config: bool,
                 click.confirm("Is this configuration correct?")
 
     chatgpt_config = dict()
-    if click.confirm("Configure ChatGPT Service?"):
+    if click.confirm("Configure ChatGPT LLM?"):
         config_confirmed = False
         while not config_confirmed:
             gpt_key = click.prompt("ChatGPT API Key", type=str)
@@ -189,7 +194,7 @@ def make_keys_config(write_config: bool,
                 click.confirm("Is this configuration correct?")
 
     fastchat_config = dict()
-    if click.confirm("Configure FastChat Service?"):
+    if click.confirm("Configure FastChat LLM?"):
         config_confirmed = False
         while not config_confirmed:
             model = click.prompt("FastChat Model", type=str,
@@ -213,11 +218,93 @@ def make_keys_config(write_config: bool,
             click.echo(pformat(fastchat_config))
             config_confirmed = click.confirm("Is this configuration correct?")
 
+    palm2_config = dict()
+    if click.confirm("Configure PaLM2 LLM?"):
+        config_confirmed = False
+        while not config_confirmed:
+            role = click.prompt("PaLM2 Role", type=str,
+                                default="You are trying to give a short "
+                                        "answer in less than 40 words.")
+            context = click.prompt("PaLM2 context depth", type=int,
+                                   default=3)
+            max_tokens = click.prompt("Max number of tokens in responses",
+                                      type=int, default=100)
+            num_processes = click.prompt(
+                "Number of queries to handle in parallel",
+                type=int, default=1)
+            palm2_config = {
+                "key_path": "/config/neon/google.json",
+                "role": role,
+                "context_depth": context,
+                "max_tokens": max_tokens,
+                "num_parallel_processes": num_processes
+            }
+            click.echo(pformat(palm2_config))
+            config_confirmed = click.confirm("Is this configuration correct?")
+
+    gemini_config = dict()
+    if click.confirm("Configure Gemini LLM?"):
+        config_confirmed = False
+        while not config_confirmed:
+            model = click.prompt("Gemini Model", type=str,
+                                 default="gemini-pro")
+            role = click.prompt("Gemini Role", type=str,
+                                default="You are trying to give a short "
+                                        "answer in less than 40 words.")
+            context = click.prompt("Gemini context depth", type=int,
+                                   default=3)
+            max_tokens = click.prompt("Max number of tokens in responses",
+                                      type=int, default=100)
+            num_processes = click.prompt(
+                "Number of queries to handle in parallel",
+                type=int, default=1)
+            gemini_config = {
+                "key_path": "/config/neon/google.json",
+                "model": model,
+                "role": role,
+                "context_depth": context,
+                "max_tokens": max_tokens,
+                "num_parallel_processes": num_processes
+            }
+            click.echo(pformat(gemini_config))
+            config_confirmed = click.confirm("Is this configuration correct?")
+
+    claude_config = dict()
+    if click.confirm("Configure Anthropic Claude LLM?"):
+        config_confirmed = False
+        while not config_confirmed:
+            anthropic_key = click.prompt("Antrhopic API Key", type=str)
+            openai_key = click.prompt("OpenAI API Key", type=str,
+                                      default=chatgpt_config.get('key'))
+            model = click.prompt("Anthropic Model", type=str,
+                                     default="claude-2")
+            role = click.prompt("Role", type=str,
+                                default="You are trying to give a short "
+                                        "answer in less than 40 words.")
+            context = click.prompt("Context depth", type=int, default=3)
+            max_tokens = click.prompt("Maximum tokens in responses", type=int,
+                                      default=256)
+            claude_config = {
+                "key": anthropic_key,
+                "openai_key": openai_key,
+                "model": model,
+                "role": role,
+                "context_depth": context,
+                "max_tokens": max_tokens
+            }
+            click.echo(pformat(claude_config))
+            config_confirmed = \
+                click.confirm("Is this configuration correct?")
+
     config = {"keys": {"api_services": api_services,
                        "emails": email_config,
                        "track_my_brands": brands_config},
               "LLM_CHAT_GPT": chatgpt_config,
-              "FastChat": fastchat_config
+              "LLM_FASTCHAT": fastchat_config,
+              "LLM_PALM2": palm2_config,
+              "LLM_GEMINI": gemini_config,
+              "LLM_CLAUDE": claude_config,
+              "FastChat": fastchat_config  # TODO: Backwards-compat. only
               }
     if write_config:
         click.echo(f"Writing configuration to {output_file}")
@@ -509,6 +596,18 @@ def configure_backend(username: str = None,
             llm_config = make_llm_bot_config()
         else:
             llm_config = dict()
+        # Check if google.json file is expected
+        if any((keys_config['LLM_PALM2'], keys_config['LLM_GEMINI'])):
+            handled = False
+            while not handled:
+                cred = click.prompt("Path to Google credential file", type=str)
+                cred = expanduser(cred)
+                if not isfile(cred):
+                    click.echo(f"Invalid path ({cred}). Please, try again.")
+                else:
+                    shutil.copyfile(cred, join(dirname(diana_config),
+                                               "google.json"))
+                    handled = True
         config = {**{"MQ": {"users": mq_auth_config,
                             "server": "neon-rabbitmq",
                             "port": 5672}},

diff --git a/neon_diana_utils/helm_charts/backend/diana-backend/Chart.yaml b/neon_diana_utils/helm_charts/backend/diana-backend/Chart.yaml
@@ -15,7 +15,7 @@ type: application
 # This is the chart version. This version number should be incremented each time you make changes
 # to the chart and its templates, including the app version.
 # Versions are expected to follow Semantic Versioning (https://semver.org/)
-version: 0.1.14
+version: 0.1.22
 
 # This is the version number of the application being deployed. This version number should be
 # incremented each time you make changes to the application. Versions are not expected to
@@ -35,5 +35,5 @@ dependencies:
     version: 0.0.11
     repository: file://../http-services
   - name: diana-mq
-    version: 0.0.9
+    version: 0.0.16
     repository: file://../mq-services
diff --git a/...iana-backend/templates/_config-secret.tpl → ...ackend/templates/_diana-config-secret.tpl b/...iana-backend/templates/_config-secret.tpl → ...ackend/templates/_diana-config-secret.tpl
diff --git a/neon_diana_utils/helm_charts/backend/diana-backend/templates/_google-config-secret.tpl b/neon_diana_utils/helm_charts/backend/diana-backend/templates/_google-config-secret.tpl
@@ -0,0 +1,10 @@
+{{- define "google_json.secret"}}
+apiVersion: v1
+kind: Secret
+metadata:
+  name: {{ tpl .Values.backend.googleSecret . }}
+type: Opaque
+data:
+  "google.json": |-
+    {{ tpl .Values.backend.googleJson . }}
+{{- end -}}
diff --git a/neon_diana_utils/helm_charts/backend/diana-backend/values.yaml b/neon_diana_utils/helm_charts/backend/diana-backend/values.yaml
@@ -6,8 +6,10 @@
 ghTokenEncoded: ''
 dianaConfig: ''
 rabbitMqConfig: ''
+googleJson: ''
 
 configSecret: diana-config
+googleSecret: google-json
 
 letsencrypt:
   sslSecret: &ssl_secret letsencrypt-private-key

diff --git a/neon_diana_utils/helm_charts/backend/mq-services/Chart.yaml b/neon_diana_utils/helm_charts/backend/mq-services/Chart.yaml
@@ -3,30 +3,46 @@ name: diana-mq
 description: Deploy DIANA MQ Services
 
 type: application
-version: 0.0.9
+version: 0.0.16
 appVersion: "1.0.1a18"
 dependencies:
   - name: neon-api-proxy
     alias: neon-api-proxy
-    version: 0.0.5
+    version: 0.0.6
     repository: file://../../mq/neon-api-proxy
   - name: neon-brands-service
     alias: neon-brands-service
-    version: 0.0.5
+    version: 0.0.6
     repository: file://../../mq/neon-brands-service
   - name: neon-email-proxy
     alias: neon-email-proxy
-    version: 0.0.5
+    version: 0.0.6
     repository: file://../../mq/neon-email-proxy
   - name: neon-metrics-service
     alias: neon-metrics-service
-    version: 0.0.6
+    version: 0.0.7
     repository: file://../../mq/neon-metrics-service
   - name: neon-script-parser
     alias: neon-script-parser
-    version: 0.0.5
+    version: 0.0.6
     repository: file://../../mq/neon-script-parser
   - name: neon-llm-chatgpt
     alias: neon-llm-chatgpt
-    version: 0.0.6
+    version: 0.0.7
     repository: file://../../mq/neon-llm-chatgpt
+  - name: neon-llm-fastchat
+    alias: neon-llm-fastchat
+    version: 0.0.6
+    repository: file://../../mq/neon-llm-fastchat
+  - name: neon-llm-claude
+    alias: neon-llm-claude
+    version: 0.0.2
+    repository: file://../../mq/neon-llm-claude
+  - name: neon-llm-gemini
+    alias: neon-llm-gemini
+    version: 0.0.2
+    repository: file://../../mq/neon-llm-gemini
+  - name: neon-llm-palm
+    alias: neon-llm-palm
+    version: 0.0.6
+    repository: file://../../mq/neon-llm-palm
diff --git a/neon_diana_utils/helm_charts/base/base-mq/Chart.yaml b/neon_diana_utils/helm_charts/base/base-mq/Chart.yaml
@@ -3,5 +3,5 @@ name: base-mq
 description: Library chart for basic MQ Services
 
 type: library
-version: 0.0.5
+version: 0.0.10
 appVersion: "1.0.1a18"
diff --git a/neon_diana_utils/helm_charts/base/base-mq/templates/_deployment.tpl b/neon_diana_utils/helm_charts/base/base-mq/templates/_deployment.tpl
@@ -41,8 +41,15 @@ spec:
         - name: config
           projected:
             sources:
+            {{- if .Values.configSecrets }}
+            {{- range .Values.configSecrets }}
+              - secret:
+                  name: {{ . }}
+            {{- end }}
+            {{ else }}
               - secret:
                   name: {{ .Values.configSecret }}
+            {{- end }}
         {{- if .Values.persistentVolumeClaim }}
         - name: {{ .Values.persistentVolumeClaim.name }}
           persistentVolumeClaim:

diff --git a/neon_diana_utils/helm_charts/mq/neon-api-proxy/Chart.yaml b/neon_diana_utils/helm_charts/mq/neon-api-proxy/Chart.yaml
@@ -3,10 +3,10 @@ name: neon-api-proxy
 description: Deploy an MQ API Proxy
 
 type: application
-version: 0.0.5
+version: 0.0.6
 appVersion: "1.0.1a18"
 
 dependencies:
   - name: base-mq
-    version: 0.0.5
+    version: 0.0.10
     repository: file://../../base/base-mq
diff --git a/neon_diana_utils/helm_charts/mq/neon-brands-service/Chart.yaml b/neon_diana_utils/helm_charts/mq/neon-brands-service/Chart.yaml
@@ -3,10 +3,10 @@ name: neon-brands-service
 description: Deploy an MQ Brands Service
 
 type: application
-version: 0.0.5
+version: 0.0.6
 appVersion: "1.0.1a18"
 
 dependencies:
   - name: base-mq
-    version: 0.0.5
+    version: 0.0.10
     repository: file://../../base/base-mq
diff --git a/neon_diana_utils/helm_charts/mq/neon-email-proxy/Chart.yaml b/neon_diana_utils/helm_charts/mq/neon-email-proxy/Chart.yaml
@@ -3,10 +3,10 @@ name: neon-email-proxy
 description: Deploy an Email Proxy Service
 
 type: application
-version: 0.0.5
+version: 0.0.6
 appVersion: "1.0.1a18"
 
 dependencies:
   - name: base-mq
-    version: 0.0.5
+    version: 0.0.10
     repository: file://../../base/base-mq
diff --git a/neon_diana_utils/helm_charts/mq/neon-llm-chatgpt/Chart.yaml b/neon_diana_utils/helm_charts/mq/neon-llm-chatgpt/Chart.yaml
@@ -3,10 +3,10 @@ name: neon-llm-chatgpt
 description: Deploy an LLM proxy for Chat GPT
 
 type: application
-version: 0.0.6
+version: 0.0.7
 appVersion: "1.0.1a18"
 
 dependencies:
   - name: base-mq
-    version: 0.0.5
+    version: 0.0.10
     repository: file://../../base/base-mq
diff --git a/neon_diana_utils/helm_charts/mq/neon-llm-claude/.helmignore b/neon_diana_utils/helm_charts/mq/neon-llm-claude/.helmignore
@@ -0,0 +1,23 @@
+# Patterns to ignore when building packages.
+# This supports shell glob matching, relative path matching, and
+# negation (prefixed with !). Only one pattern per line.
+.DS_Store
+# Common VCS dirs
+.git/
+.gitignore
+.bzr/
+.bzrignore
+.hg/
+.hgignore
+.svn/
+# Common backup files
+*.swp
+*.bak
+*.tmp
+*.orig
+*~
+# Various IDEs
+.project
+.idea/
+*.tmproj
+.vscode/
diff --git a/neon_diana_utils/helm_charts/mq/neon-llm-claude/Chart.yaml b/neon_diana_utils/helm_charts/mq/neon-llm-claude/Chart.yaml
@@ -0,0 +1,12 @@
+apiVersion: v2
+name: neon-llm-claude
+description: Deploy an LLM proxy for Claude by Anthropic
+
+type: application
+version: 0.0.2
+appVersion: "1.0.1a16"
+
+dependencies:
+  - name: base-mq
+    version: 0.0.10
+    repository: file://../../base/base-mq