Merge pull request #111 from doringeman/misc

Use model structs imported from model-runner

Author: doringeman

commands/compose.go

@@ -12,6 +12,7 @@ import (
 	"github.com/docker/model-cli/desktop"
 	"github.com/docker/model-runner/pkg/inference/backends/llamacpp"
 	"github.com/docker/model-runner/pkg/inference/scheduling"
+	dmrm "github.com/docker/model-runner/pkg/inference/models"
 	"github.com/spf13/cobra"
 )
 
@@ -143,7 +144,7 @@ func downloadModelsOnlyIfNotFound(desktopClient *desktop.Client, models []string
 	}
 	for _, model := range models {
 		// Download the model if not already present in the local model store
-		if !slices.ContainsFunc(modelsDownloaded, func(m desktop.Model) bool {
+		if !slices.ContainsFunc(modelsDownloaded, func(m dmrm.Model) bool {
 			if model == m.ID {
 				return true
 			}

commands/list.go

@@ -3,15 +3,17 @@ package commands
 import (
 	"bytes"
 	"fmt"
+	"os"
+	"time"
+
 	"github.com/docker/go-units"
 	"github.com/docker/model-cli/commands/completion"
 	"github.com/docker/model-cli/commands/formatter"
 	"github.com/docker/model-cli/desktop"
 	"github.com/docker/model-cli/pkg/standalone"
+	dmrm "github.com/docker/model-runner/pkg/inference/models"
 	"github.com/olekukonko/tablewriter"
 	"github.com/spf13/cobra"
-	"os"
-	"time"
 )
 
 func newListCmd() *cobra.Command {
@@ -79,7 +81,7 @@ func listModels(openai bool, desktopClient *desktop.Client, quiet bool, jsonForm
 	return prettyPrintModels(models), nil
 }
 
-func prettyPrintModels(models []desktop.Model) string {
+func prettyPrintModels(models []dmrm.Model) string {
 	var buf bytes.Buffer
 	table := tablewriter.NewWriter(&buf)
 
@@ -116,7 +118,7 @@ func prettyPrintModels(models []desktop.Model) string {
 	return buf.String()
 }
 
-func appendRow(table *tablewriter.Table, tag string, model desktop.Model) {
+func appendRow(table *tablewriter.Table, tag string, model dmrm.Model) {
 	if len(model.ID) < 19 {
 		fmt.Fprintf(os.Stderr, "invalid model ID for model: %v\n", model)
 		return

desktop/api.go

@@ -31,39 +31,3 @@ type OpenAIChatResponse struct {
 		FinishReason string `json:"finish_reason"`
 	} `json:"choices"`
 }
-
-type OpenAIModel struct {
-	ID      string `json:"id"`
-	Object  string `json:"object"`
-	Created int64  `json:"created"`
-	OwnedBy string `json:"owned_by"`
-}
-
-type OpenAIModelList struct {
-	Object string         `json:"object"`
-	Data   []*OpenAIModel `json:"data"`
-}
-
-// TODO: To be replaced by the Model struct from pianta's common/pkg/inference/models/api.go.
-// (https://github.com/docker/pinata/pull/33331)
-type Format string
-
-type Config struct {
-	Format       Format  `json:"format,omitempty"`
-	Quantization string  `json:"quantization,omitempty"`
-	Parameters   string  `json:"parameters,omitempty"`
-	Architecture string  `json:"architecture,omitempty"`
-	Size         string  `json:"size,omitempty"`
-	ContextSize  uint64 `json:"context_size,omitzero"`
-}
-
-type Model struct {
-	// ID is the globally unique model identifier.
-	ID string `json:"id"`
-	// Tags are the list of tags associated with the model.
-	Tags []string `json:"tags,omitempty"`
-	// Created is the Unix epoch timestamp corresponding to the model creation.
-	Created int64 `json:"created"`
-	// Config describes the model.
-	Config Config `json:"config"`
-}

desktop/desktop.go

@@ -13,7 +13,7 @@ import (
 	"time"
 
 	"github.com/docker/model-runner/pkg/inference"
-	"github.com/docker/model-runner/pkg/inference/models"
+	dmrm "github.com/docker/model-runner/pkg/inference/models"
 	"github.com/docker/model-runner/pkg/inference/scheduling"
 	"github.com/pkg/errors"
 	"go.opentelemetry.io/otel"
@@ -102,7 +102,7 @@ func (c *Client) Status() Status {
 
 func (c *Client) Pull(model string, progress func(string)) (string, bool, error) {
 	model = normalizeHuggingFaceModelName(model)
-	jsonData, err := json.Marshal(models.ModelCreateRequest{From: model})
+	jsonData, err := json.Marshal(dmrm.ModelCreateRequest{From: model})
 	if err != nil {
 		return "", false, fmt.Errorf("error marshaling request: %w", err)
 	}
@@ -207,73 +207,73 @@ func (c *Client) Push(model string, progress func(string)) (string, bool, error)
 	return "", progressShown, fmt.Errorf("unexpected end of stream while pushing model %s", model)
 }
 
-func (c *Client) List() ([]Model, error) {
+func (c *Client) List() ([]dmrm.Model, error) {
 	modelsRoute := inference.ModelsPrefix
 	body, err := c.listRaw(modelsRoute, "")
 	if err != nil {
-		return []Model{}, err
+		return []dmrm.Model{}, err
 	}
 
-	var modelsJson []Model
+	var modelsJson []dmrm.Model
 	if err := json.Unmarshal(body, &modelsJson); err != nil {
 		return modelsJson, fmt.Errorf("failed to unmarshal response body: %w", err)
 	}
 
 	return modelsJson, nil
 }
 
-func (c *Client) ListOpenAI() (OpenAIModelList, error) {
+func (c *Client) ListOpenAI() (dmrm.OpenAIModelList, error) {
 	modelsRoute := inference.InferencePrefix + "/v1/models"
 	rawResponse, err := c.listRaw(modelsRoute, "")
 	if err != nil {
-		return OpenAIModelList{}, err
+		return dmrm.OpenAIModelList{}, err
 	}
-	var modelsJson OpenAIModelList
+	var modelsJson dmrm.OpenAIModelList
 	if err := json.Unmarshal(rawResponse, &modelsJson); err != nil {
 		return modelsJson, fmt.Errorf("failed to unmarshal response body: %w", err)
 	}
 	return modelsJson, nil
 }
 
-func (c *Client) Inspect(model string, remote bool) (Model, error) {
+func (c *Client) Inspect(model string, remote bool) (dmrm.Model, error) {
 	model = normalizeHuggingFaceModelName(model)
 	if model != "" {
 		if !strings.Contains(strings.Trim(model, "/"), "/") {
 			// Do an extra API call to check if the model parameter isn't a model ID.
 			modelId, err := c.fullModelID(model)
 			if err != nil {
-				return Model{}, fmt.Errorf("invalid model name: %s", model)
+				return dmrm.Model{}, fmt.Errorf("invalid model name: %s", model)
 			}
 			model = modelId
 		}
 	}
 	rawResponse, err := c.listRawWithQuery(fmt.Sprintf("%s/%s", inference.ModelsPrefix, model), model, remote)
 	if err != nil {
-		return Model{}, err
+		return dmrm.Model{}, err
 	}
-	var modelInspect Model
+	var modelInspect dmrm.Model
 	if err := json.Unmarshal(rawResponse, &modelInspect); err != nil {
 		return modelInspect, fmt.Errorf("failed to unmarshal response body: %w", err)
 	}
 
 	return modelInspect, nil
 }
 
-func (c *Client) InspectOpenAI(model string) (OpenAIModel, error) {
+func (c *Client) InspectOpenAI(model string) (dmrm.OpenAIModel, error) {
 	model = normalizeHuggingFaceModelName(model)
 	modelsRoute := inference.InferencePrefix + "/v1/models"
 	if !strings.Contains(strings.Trim(model, "/"), "/") {
 		// Do an extra API call to check if the model parameter isn't a model ID.
 		var err error
 		if model, err = c.fullModelID(model); err != nil {
-			return OpenAIModel{}, fmt.Errorf("invalid model name: %s", model)
+			return dmrm.OpenAIModel{}, fmt.Errorf("invalid model name: %s", model)
 		}
 	}
 	rawResponse, err := c.listRaw(fmt.Sprintf("%s/%s", modelsRoute, model), model)
 	if err != nil {
-		return OpenAIModel{}, err
+		return dmrm.OpenAIModel{}, err
 	}
-	var modelInspect OpenAIModel
+	var modelInspect dmrm.OpenAIModel
 	if err := json.Unmarshal(rawResponse, &modelInspect); err != nil {
 		return modelInspect, fmt.Errorf("failed to unmarshal response body: %w", err)
 	}
@@ -315,7 +315,7 @@ func (c *Client) fullModelID(id string) (string, error) {
 		return "", err
 	}
 
-	var modelsJson []Model
+	var modelsJson []dmrm.Model
 	if err := json.Unmarshal(bodyResponse, &modelsJson); err != nil {
 		return "", fmt.Errorf("failed to unmarshal response body: %w", err)
 	}