chore: default to gemma-3-12b-it-qat (#60)

* chore: default to gemma-3-12b-it-qat

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

* fix: simplify tests to run faster

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

---------

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
This commit is contained in:
Ettore Di Giacinto
2025-04-22 11:44:42 +02:00
committed by GitHub
parent ee1667d51a
commit c4618896cf
4 changed files with 16 additions and 10 deletions

View File

@@ -9,7 +9,7 @@ cleanup-tests:
docker compose down docker compose down
tests: prepare-tests tests: prepare-tests
LOCALAGI_MODEL="arcee-agent" LOCALAI_API_URL="http://localhost:8081" LOCALAGI_API_URL="http://localhost:8080" $(GOCMD) run github.com/onsi/ginkgo/v2/ginkgo --fail-fast -v -r ./... LOCALAGI_MODEL="gemma-3-12b-it-qat" LOCALAI_API_URL="http://localhost:8081" LOCALAGI_API_URL="http://localhost:8080" $(GOCMD) run github.com/onsi/ginkgo/v2/ginkgo --fail-fast -v -r ./...
run-nokb: run-nokb:
$(MAKE) run KBDISABLEINDEX=true $(MAKE) run KBDISABLEINDEX=true

View File

@@ -114,7 +114,7 @@ LocalAGI supports multiple hardware configurations through Docker Compose profil
- Supports text, multimodal, and image generation models - Supports text, multimodal, and image generation models
- Run with: `docker compose -f docker-compose.nvidia.yaml up` - Run with: `docker compose -f docker-compose.nvidia.yaml up`
- Default models: - Default models:
- Text: `arcee-agent` - Text: `gemma-3-12b-it-qat`
- Multimodal: `minicpm-v-2_6` - Multimodal: `minicpm-v-2_6`
- Image: `sd-1.5-ggml` - Image: `sd-1.5-ggml`
- Environment variables: - Environment variables:
@@ -130,7 +130,7 @@ LocalAGI supports multiple hardware configurations through Docker Compose profil
- Supports text, multimodal, and image generation models - Supports text, multimodal, and image generation models
- Run with: `docker compose -f docker-compose.intel.yaml up` - Run with: `docker compose -f docker-compose.intel.yaml up`
- Default models: - Default models:
- Text: `arcee-agent` - Text: `gemma-3-12b-it-qat`
- Multimodal: `minicpm-v-2_6` - Multimodal: `minicpm-v-2_6`
- Image: `sd-1.5-ggml` - Image: `sd-1.5-ggml`
- Environment variables: - Environment variables:
@@ -161,7 +161,7 @@ docker compose -f docker-compose.intel.yaml up
``` ```
If no models are specified, it will use the defaults: If no models are specified, it will use the defaults:
- Text model: `arcee-agent` - Text model: `gemma-3-12b-it-qat`
- Multimodal model: `minicpm-v-2_6` - Multimodal model: `minicpm-v-2_6`
- Image model: `sd-1.5-ggml` - Image model: `sd-1.5-ggml`

View File

@@ -226,7 +226,10 @@ var _ = Describe("Agent test", func() {
WithLLMAPIKey(apiKeyURL), WithLLMAPIKey(apiKeyURL),
WithTimeout("10m"), WithTimeout("10m"),
WithActions( WithActions(
actions.NewSearch(map[string]string{}), &TestAction{response: map[string]string{
"boston": testActionResult,
"milan": testActionResult2,
}},
), ),
EnablePlanning, EnablePlanning,
EnableForceReasoning, EnableForceReasoning,
@@ -238,18 +241,21 @@ var _ = Describe("Agent test", func() {
defer agent.Stop() defer agent.Stop()
result := agent.Ask( result := agent.Ask(
types.WithText("Thoroughly plan a trip to San Francisco from Venice, Italy; check flight times, visa requirements and whether electrical items are allowed in cabin luggage."), types.WithText("Use the plan tool to do two actions in sequence: search for the weather in boston and search for the weather in milan"),
) )
Expect(len(result.State)).To(BeNumerically(">", 1)) Expect(len(result.State)).To(BeNumerically(">", 1))
actionsExecuted := []string{} actionsExecuted := []string{}
actionResults := []string{}
for _, r := range result.State { for _, r := range result.State {
xlog.Info(r.Result) xlog.Info(r.Result)
actionsExecuted = append(actionsExecuted, r.Action.Definition().Name.String()) actionsExecuted = append(actionsExecuted, r.Action.Definition().Name.String())
actionResults = append(actionResults, r.ActionResult.Result)
} }
Expect(actionsExecuted).To(ContainElement("search_internet"), fmt.Sprint(result)) Expect(actionsExecuted).To(ContainElement("get_weather"), fmt.Sprint(result))
Expect(actionsExecuted).To(ContainElement("plan"), fmt.Sprint(result)) Expect(actionsExecuted).To(ContainElement("plan"), fmt.Sprint(result))
Expect(actionResults).To(ContainElement(testActionResult), fmt.Sprint(result))
Expect(actionResults).To(ContainElement(testActionResult2), fmt.Sprint(result))
}) })
It("Can initiate conversations", func() { It("Can initiate conversations", func() {

View File

@@ -7,7 +7,7 @@ services:
# Image list (dockerhub): https://hub.docker.com/r/localai/localai # Image list (dockerhub): https://hub.docker.com/r/localai/localai
image: localai/localai:master-ffmpeg-core image: localai/localai:master-ffmpeg-core
command: command:
- ${MODEL_NAME:-arcee-agent} - ${MODEL_NAME:-gemma-3-12b-it-qat}
- ${MULTIMODAL_MODEL:-minicpm-v-2_6} - ${MULTIMODAL_MODEL:-minicpm-v-2_6}
- ${IMAGE_MODEL:-sd-1.5-ggml} - ${IMAGE_MODEL:-sd-1.5-ggml}
- granite-embedding-107m-multilingual - granite-embedding-107m-multilingual
@@ -59,7 +59,7 @@ services:
- 8080:3000 - 8080:3000
#image: quay.io/mudler/localagi:master #image: quay.io/mudler/localagi:master
environment: environment:
- LOCALAGI_MODEL=${MODEL_NAME:-arcee-agent} - LOCALAGI_MODEL=${MODEL_NAME:-gemma-3-12b-it-qat}
- LOCALAGI_MULTIMODAL_MODEL=${MULTIMODAL_MODEL:-minicpm-v-2_6} - LOCALAGI_MULTIMODAL_MODEL=${MULTIMODAL_MODEL:-minicpm-v-2_6}
- LOCALAGI_IMAGE_MODEL=${IMAGE_MODEL:-sd-1.5-ggml} - LOCALAGI_IMAGE_MODEL=${IMAGE_MODEL:-sd-1.5-ggml}
- LOCALAGI_LLM_API_URL=http://localai:8080 - LOCALAGI_LLM_API_URL=http://localai:8080