add scripts

.gitignore

@@ -24,3 +24,4 @@ venv/
 pytest-report.xml
 .coverage
 .python-version
+data

kvant_build_local.sh

@@ -0,0 +1,6 @@
+#!/usr/bin/env bash
+
+export USE_COPY_NOT_MOUNT=true
+export LLAMA_STACK_DIR=.
+
+uvx --from . llama stack build --template kvant --image-type container  --image-name kvant

kvant_start_local.sh

@@ -0,0 +1,25 @@
+#!/usr/bin/env bash
+
+export INFERENCE_MODEL="inference-llama4-maverick"
+export EMBEDDING_MODEL="inference-bge-m3"
+export EMBEDDING_DIMENSION="1024"
+export LLAMA_STACK_PORT=8321
+export OPENAI_BASE_URL=https://maas.ai-2.kvant.cloud/v1
+export OPENAI_API_KEY=sk-ZqAWqBKFXjb6y3tVej2AaA
+export VLLM_MAX_TOKENS=125000
+
+docker run -it \
+  -p $LLAMA_STACK_PORT:$LLAMA_STACK_PORT \
+  -v $(pwd)/data:/root/.llama \
+  --mount type=bind,source="$(pwd)"/llama_stack/templates/kvant/run.yaml,target=/root/.llama/config.yaml,readonly \
+  --entrypoint python \
+  distribution-kvant:dev \
+  -m llama_stack.distribution.server.server  --config /root/.llama/config.yaml \
+  --port $LLAMA_STACK_PORT \
+  --env VLLM_URL=$OPENAI_BASE_URL \
+  --env VLLM_API_TOKEN=$OPENAI_API_KEY \
+  --env PASSTHROUGH_URL=$OPENAI_BASE_URL \
+  --env PASSTHROUGH_API_KEY=$OPENAI_API_KEY \
+  --env INFERENCE_MODEL=$INFERENCE_MODEL \
+  --env EMBEDDING_MODEL=$EMBEDDING_MODEL \
+  --env EMBEDDING_DIMENSION=$EMBEDDING_DIMENSION \

llama_stack/templates/kvant/run.yaml

@@ -30,9 +30,6 @@ providers:
         type: sqlite
         namespace: null
         db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/kvant}/faiss_store.db
-      responses_store:
-        type: sqlite
-        db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/bedrock}/responses_store.db
   safety:
   - provider_id: llama-guard
     provider_type: inline::llama-guard
@@ -46,6 +43,9 @@ providers:
         type: sqlite
         namespace: null
         db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/kvant}/agents_store.db
+      responses_store:
+        type: sqlite
+        db_path: ${env.SQLITE_STORE_DIR:~/.llama/distributions/kvant}/responses_store.db
   telemetry:
   - provider_id: meta-reference
     provider_type: inline::meta-reference

playground_start_local.sh

@@ -0,0 +1,3 @@
+#!/usr/bin/env bash
+
+uv run --with ".[ui]" streamlit run llama_stack/distribution/ui/app.py