fix config hot-reload on macos (#180)

Co-authored-by: srevn <srevn@github>

.goreleaser.yaml

@@ -17,14 +17,16 @@ builds:
       - goos: windows
         goarch: arm64
 
-# use zip format for windows
 archives:
   - id: default
-    format: tar.gz
+    formats:
+      - tar.gz
     name_template: "{{ .ProjectName }}_{{ .Version }}_{{ .Os }}_{{ .Arch }}"
     builds_info:
       group: root
       owner: root
     format_overrides:
+      # use zip format for windows
       - goos: windows
-        format: zip
+        formats:
+          - zip

README.md

@@ -22,6 +22,7 @@ Written in golang, it is very easy to install (single binary with no dependencie
   - `v1/audio/speech` ([#36](https://github.com/mostlygeek/llama-swap/issues/36))
   - `v1/audio/transcriptions` ([docs](https://github.com/mostlygeek/llama-swap/issues/41#issuecomment-2722637867))
 - ✅ llama-swap custom API endpoints
+  - `/ui` - web UI
   - `/log` - remote log monitoring
   - `/upstream/:model_id` - direct access to upstream HTTP server ([demo](https://github.com/mostlygeek/llama-swap/pull/31))
   - `/unload` - manually unload running models ([#58](https://github.com/mostlygeek/llama-swap/issues/58))
@@ -67,6 +68,12 @@ However, there are many more capabilities that llama-swap supports:
 
 See the [configuration documentation](https://github.com/mostlygeek/llama-swap/wiki/Configuration) in the wiki all options and examples.
 
+## Web UI
+
+llama-swap ships with a web based interface to make it easier to monitor logs and check the status of models. 
+
+<img width="1758" alt="image" src="https://github.com/user-attachments/assets/31ae5bcd-5efd-46b0-b64b-6db9e60196d3" />
+
 ## Docker Install ([download images](https://github.com/mostlygeek/llama-swap/pkgs/container/llama-swap))
 
 Docker is the quickest way to try out llama-swap:

config.example.yaml

@@ -1,93 +1,191 @@
-# ======
-# For a more detailed configuration example:
-# https://github.com/mostlygeek/llama-swap/wiki/Configuration
-# ======
+# llama-swap YAML configuration example
+# -------------------------------------
+#
+# - Below are all the available configuration options for llama-swap.
+# - Settings with a default value, or noted as optional can be omitted.
+# - Settings that are marked required must be in your configuration file
 
-# Seconds to wait for llama.cpp to be available to serve requests
-# Default (and minimum): 15 seconds
-healthCheckTimeout: 90
+# healthCheckTimeout: number of seconds to wait for a model to be ready to serve requests
+# - optional, default: 120
+# - minimum value is 15 seconds, anything less will be set to this value
+healthCheckTimeout: 500
 
-# valid log levels: debug, info (default), warn, error
-logLevel: debug
+# logLevel: sets the logging value
+# - optional, default: info
+# - Valid log levels: debug, info, warn, error
+logLevel: info
 
-# creating a coding profile with models for code generation and general questions
-groups:
-  coding:
-    swap: false
-    members:
-      - "qwen"
-      - "llama"
+# startPort: sets the starting port number for the automatic ${PORT} macro.
+# - optional, default: 5800
+# - the ${PORT} macro can be used in model.cmd and model.proxy settings
+# - it is automatically incremented for every model that uses it
+startPort: 10001
 
+# macros: sets a dictionary of string:string pairs
+# - optional, default: empty dictionary
+# - these are reusable snippets
+# - used in a model's cmd, cmdStop, proxy and checkEndpoint
+# - useful for reducing common configuration settings
+macros:
+  "latest-llama": >
+    /path/to/llama-server/llama-server-ec9e0301
+    --port ${PORT}
+
+# models: a dictionary of model configurations
+# - required
+# - each key is the model's ID, used in API requests
+# - model settings have default values that are used if they are not defined here
+# - below are examples of the various settings a model can have:
+# - available model settings: env, cmd, cmdStop, proxy, aliases, checkEndpoint, ttl, unlisted
 models:
+
+  # keys are the model names used in API requests
   "llama":
+    # cmd: the command to run to start the inference server.
+    # - required
+    # - it is just a string, similar to what you would run on the CLI
+    # - using `|` allows for comments in the command, these will be parsed out
+    # - macros can be used within cmd
     cmd: |
-      models/llama-server-osx
-      --port ${PORT}
-      -m models/Llama-3.2-1B-Instruct-Q4_0.gguf
+      # ${latest-llama} is a macro that is defined above
+      ${latest-llama}
+      --model path/to/Qwen2.5-1.5B-Instruct-Q4_K_M.gguf
 
-    # list of model name aliases this llama.cpp instance can serve
+    # env: define an array of environment variables to inject into cmd's environment
+    # - optional, default: empty array
+    # - each value is a single string
+    # - in the format: ENV_NAME=value
+    env:
+      - "CUDA_VISIBLE_DEVICES=0,1,2"
+
+    # proxy: the URL where llama-swap routes API requests
+    # - optional, default: http://localhost:${PORT}
+    # - if you used ${PORT} in cmd this can be omitted
+    # - if you use a custom port in cmd this *must* be set
+    proxy: http://127.0.0.1:8999
+
+    # aliases: alternative model names that this model configuration is used for
+    # - optional, default: empty array
+    # - aliases must be unique globally
+    # - useful for impersonating a specific model
     aliases:
-    - gpt-4o-mini
+      - "gpt-4o-mini"
+      - "gpt-3.5-turbo"
 
-    # check this path for a HTTP 200 response for the server to be ready
-    checkEndpoint: /health
+    # checkEndpoint: URL path to check if the server is ready
+    # - optional, default: /health
+    # - use "none" to skip endpoint ready checking
+    # - endpoint is expected to return an HTTP 200 response
+    # - all requests wait until the endpoint is ready (or fails)
+    checkEndpoint: /custom-endpoint
 
-    # unload model after 5 seconds
-    ttl: 5
+    # ttl: automatically unload the model after this many seconds
+    # - optional, default: 0
+    # - ttl values must be a value greater than 0
+    # - a value of 0 disables automatic unloading of the model
+    ttl: 60
 
-  "qwen":
-    cmd: models/llama-server-osx --port ${PORT} -m models/qwen2.5-0.5b-instruct-q8_0.gguf
-    aliases:
-      - gpt-3.5-turbo
+    # useModelName: overrides the model name that is sent to upstream server
+    # - optional, default: ""
+    # - useful when the upstream server expects a specific model name or format
+    useModelName: "qwen:qwq"
 
-  # Embedding example with Nomic
-  # https://huggingface.co/nomic-ai/nomic-embed-text-v1.5-GGUF
-  "nomic":
-    cmd: |
-      models/llama-server-osx --port ${PORT}
-      -m models/nomic-embed-text-v1.5.Q8_0.gguf
-      --ctx-size 8192
-      --batch-size 8192
-      --rope-scaling yarn
-      --rope-freq-scale 0.75
-      -ngl 99
-      --embeddings
+    # filters: a dictionary of filter settings
+    # - optional, default: empty dictionary
+    filters:
+      # strip_params: a comma separated list of parameters to remove from the request
+      # - optional, default: ""
+      # - useful for preventing overriding of default server params by requests
+      # - `model` parameter is never removed
+      # - can be any JSON key in the request body
+      # - recommended to stick to sampling parameters
+      strip_params: "temperature, top_p, top_k"
 
-  # Reranking example with bge-reranker
-  # https://huggingface.co/gpustack/bge-reranker-v2-m3-GGUF
-  "bge-reranker":
-    cmd: |
-      models/llama-server-osx --port ${PORT}
-      -m models/bge-reranker-v2-m3-Q4_K_M.gguf
-      --ctx-size 8192
-      --reranking
+  # Unlisted model example:
+  "qwen-unlisted":
+    # unlisted: true or false
+    # - optional, default: false
+    # - unlisted models do not show up in /v1/models or /upstream lists
+    # - can be requested as normal through all apis
+    unlisted: true
+    cmd: llama-server --port ${PORT} -m Llama-3.2-1B-Instruct-Q4_K_M.gguf -ngl 0
 
-  # Docker Support (v26.1.4+ required!)
-  "dockertest":
+  # Docker example:
+  # container run times like Docker and Podman can also be used with a
+  # a combination of cmd and cmdStop.
+  "docker-llama":
+    proxy: "http://127.0.0.1:${PORT}"
     cmd: |
       docker run --name dockertest
       --init --rm -p ${PORT}:8080 -v /mnt/nvme/models:/models
-      ghcr.io/ggerganov/llama.cpp:server
+      ghcr.io/ggml-org/llama.cpp:server
       --model '/models/Qwen2.5-Coder-0.5B-Instruct-Q4_K_M.gguf'
 
-  "simple":
-    # example of setting environment variables
-    env:
-      - CUDA_VISIBLE_DEVICES=0,1
-      - env1=hello
-    cmd: build/simple-responder --port ${PORT}
-    unlisted: true
+    # cmdStop: command to run to stop the model gracefully
+    # - optional, default: ""
+    # - useful for stopping commands managed by another system
+    # - on POSIX systems: a SIGTERM is sent for graceful shutdown
+    # - on Windows, taskkill is used
+    # - processes are given 5 seconds to shutdown until they are forcefully killed
+    # - the upstream's process id is available in the ${PID} macro
+    cmdStop: docker stop dockertest
 
-    # use "none" to skip check. Caution this may cause some requests to fail
-    # until the upstream server is ready for traffic
-    checkEndpoint: none
+# groups: a dictionary of group settings
+# - optional, default: empty dictionary
+# - provide advanced controls over model swapping behaviour.
+# - Using groups some models can be kept loaded indefinitely, while others are swapped out.
+# - model ids must be defined in the Models section
+# - a model can only be a member of one group
+# - group behaviour is controlled via the `swap`, `exclusive` and `persistent` fields
+# - see issue #109 for details
+#
+# NOTE: the example below uses model names that are not defined above for demonstration purposes
+groups:
+  # group1 is same as the default behaviour of llama-swap where only one model is allowed
+  # to run a time across the whole llama-swap instance
+  "group1":
+    # swap: controls the model swapping behaviour in within the group
+    # - optional, default: true
+    # - true : only one model is allowed to run at a time
+    # - false: all models can run together, no swapping
+    swap: true
 
-  # don't use these, just for testing if things are broken
-  "broken":
-    cmd: models/llama-server-osx --port 8999 -m models/doesnotexist.gguf
-    proxy: http://127.0.0.1:8999
-    unlisted: true
-  "broken_timeout":
-    cmd: models/llama-server-osx --port 8999 -m models/qwen2.5-0.5b-instruct-q8_0.gguf
-    proxy: http://127.0.0.1:9000
-    unlisted: true
+    # exclusive: controls how the group affects other groups
+    # - optional, default: true
+    # - true: causes all other groups to unload when this group runs a model
+    # - false: does not affect other groups
+    exclusive: true
+
+    # members references the models defined above
+    # required
+    members:
+      - "llama"
+      - "qwen-unlisted"
+
+  # Example:
+  # - in this group all the models can run at the same time
+  # - when a different group loads all running models in this group are unloaded
+  "group2":
+    swap: false
+    exclusive: false
+    members:
+      - "docker-llama"
+      - "modelA"
+      - "modelB"
+
+  # Example:
+  # - a persistent group, prevents other groups from unloading it
+  "forever":
+    # persistent: prevents over groups from unloading the models in this group
+    # - optional, default: false
+    # - does not affect individual model behaviour
+    persistent: true
+
+    # set swap/exclusive to false to prevent swapping inside the group
+    # and the unloading of other groups
+    swap: false
+    exclusive: false
+    members:
+      - "forever-modelA"
+      - "forever-modelB"
+      - "forever-modelc"

llama-swap.go

@@ -144,8 +144,8 @@ func watchConfigFileWithReload(configPath string, reloadChan chan<- *proxy.Proxy
 			if !ok {
 				return
 			}
-			// We only care about writes to the specific config file
-			if event.Name == configPath && event.Has(fsnotify.Write) {
+			// We only care about writes/creates to the specific config file
+			if event.Name == configPath && (event.Has(fsnotify.Write) || event.Has(fsnotify.Create)) {
 				// Reset or start the debounce timer
 				if debounceTimer != nil {
 					debounceTimer.Stop()

misc/simple-responder/simple-responder.go

@@ -42,9 +42,12 @@ func main() {
 			time.Sleep(wait)
 		}
 
+		bodyBytes, _ := io.ReadAll(c.Request.Body)
+
 		c.JSON(http.StatusOK, gin.H{
 			"responseMessage":  *responseMessage,
 			"h_content_length": c.Request.Header.Get("Content-Length"),
+			"request_body":     string(bodyBytes),
 		})
 	})
 

proxy/config.go

@@ -6,6 +6,7 @@ import (
 	"os"
 	"regexp"
 	"runtime"
+	"slices"
 	"sort"
 	"strconv"
 	"strings"
@@ -29,6 +30,9 @@ type ModelConfig struct {
 
 	// Limit concurrency of HTTP requests to process
 	ConcurrencyLimit int `yaml:"concurrencyLimit"`
+
+	// Model filters see issue #174
+	Filters ModelFilters `yaml:"filters"`
 }
 
 func (m *ModelConfig) UnmarshalYAML(unmarshal func(interface{}) error) error {
@@ -63,6 +67,46 @@ func (m *ModelConfig) SanitizedCommand() ([]string, error) {
 	return SanitizeCommand(m.Cmd)
 }
 
+// ModelFilters see issue #174
+type ModelFilters struct {
+	StripParams string `yaml:"strip_params"`
+}
+
+func (m *ModelFilters) UnmarshalYAML(unmarshal func(interface{}) error) error {
+	type rawModelFilters ModelFilters
+	defaults := rawModelFilters{
+		StripParams: "",
+	}
+
+	if err := unmarshal(&defaults); err != nil {
+		return err
+	}
+
+	*m = ModelFilters(defaults)
+	return nil
+}
+
+func (f ModelFilters) SanitizedStripParams() ([]string, error) {
+	if f.StripParams == "" {
+		return nil, nil
+	}
+
+	params := strings.Split(f.StripParams, ",")
+	cleaned := make([]string, 0, len(params))
+
+	for _, param := range params {
+		trimmed := strings.TrimSpace(param)
+		if trimmed == "model" || trimmed == "" {
+			continue
+		}
+		cleaned = append(cleaned, trimmed)
+	}
+
+	// sort cleaned
+	slices.Sort(cleaned)
+	return cleaned, nil
+}
+
 type GroupConfig struct {
 	Swap       bool     `yaml:"swap"`
 	Exclusive  bool     `yaml:"exclusive"`
@@ -212,6 +256,7 @@ func LoadConfigFromReader(r io.Reader) (Config, error) {
 			modelConfig.CmdStop = strings.ReplaceAll(modelConfig.CmdStop, macroSlug, macroValue)
 			modelConfig.Proxy = strings.ReplaceAll(modelConfig.Proxy, macroSlug, macroValue)
 			modelConfig.CheckEndpoint = strings.ReplaceAll(modelConfig.CheckEndpoint, macroSlug, macroValue)
+			modelConfig.Filters.StripParams = strings.ReplaceAll(modelConfig.Filters.StripParams, macroSlug, macroValue)
 		}
 
 		// enforce ${PORT} used in both cmd and proxy

proxy/config_posix_test.go

@@ -83,6 +83,9 @@ models:
 		assert.Equal(t, "", model1.UseModelName)
 		assert.Equal(t, 0, model1.ConcurrencyLimit)
 	}
+
+	// default empty filter exists
+	assert.Equal(t, "", model1.Filters.StripParams)
 }
 
 func TestConfig_LoadPosix(t *testing.T) {

proxy/config_test.go

@@ -300,3 +300,28 @@ models:
 		})
 	}
 }
+
+func TestConfig_ModelFilters(t *testing.T) {
+	content := `
+macros:
+  default_strip: "temperature, top_p"
+models:
+  model1:
+    cmd: path/to/cmd --port ${PORT}
+    filters:
+      strip_params: "model, top_k, ${default_strip}, , ,"
+`
+	config, err := LoadConfigFromReader(strings.NewReader(content))
+	assert.NoError(t, err)
+	modelConfig, ok := config.Models["model1"]
+	if !assert.True(t, ok) {
+		t.FailNow()
+	}
+
+	// make sure `model` and enmpty strings are not in the list
+	assert.Equal(t, "model, top_k, temperature, top_p, , ,", modelConfig.Filters.StripParams)
+	sanitized, err := modelConfig.Filters.SanitizedStripParams()
+	if assert.NoError(t, err) {
+		assert.Equal(t, []string{"temperature", "top_k", "top_p"}, sanitized)
+	}
+}

proxy/config_windows_test.go

@@ -80,6 +80,9 @@ models:
 		assert.Equal(t, "", model1.UseModelName)
 		assert.Equal(t, 0, model1.ConcurrencyLimit)
 	}
+
+	// default empty filter exists
+	assert.Equal(t, "", model1.Filters.StripParams)
 }
 
 func TestConfig_LoadWindows(t *testing.T) {

proxy/process.go

@@ -189,18 +189,19 @@ func (p *Process) start() error {
 	p.waitStarting.Add(1)
 	defer p.waitStarting.Done()
 	cmdContext, ctxCancelUpstream := context.WithCancel(context.Background())
-	p.proxyLogger.Debugf("<%s> Executing start command: %s", p.ID, strings.Join(args, " "))
+
 	p.cmd = exec.CommandContext(cmdContext, args[0], args[1:]...)
 	p.cmd.Stdout = p.processLogger
 	p.cmd.Stderr = p.processLogger
-	p.cmd.Env = p.config.Env
-
+	p.cmd.Env = append(p.cmd.Environ(), p.config.Env...)
 	p.cmd.Cancel = p.cmdStopUpstreamProcess
 	p.cmd.WaitDelay = p.gracefulStopTimeout
 	p.cancelUpstream = ctxCancelUpstream
 	p.cmdWaitChan = make(chan struct{})
 
 	p.failedStartCount++ // this will be reset to zero when the process has successfully started
+
+	p.proxyLogger.Debugf("<%s> Executing start command: %s, env: %s", p.ID, strings.Join(args, " "), strings.Join(p.config.Env, ", "))
 	err = p.cmd.Start()
 
 	// Set process state to failed
@@ -531,7 +532,7 @@ func (p *Process) cmdStopUpstreamProcess() error {
 		stopCmd := exec.Command(stopArgs[0], stopArgs[1:]...)
 		stopCmd.Stdout = p.processLogger
 		stopCmd.Stderr = p.processLogger
-		stopCmd.Env = p.config.Env
+		stopCmd.Env = p.cmd.Env
 
 		if err := stopCmd.Run(); err != nil {
 			p.proxyLogger.Errorf("<%s> Failed to exec stop command: %v", p.ID, err)

proxy/process_test.go

@@ -394,6 +394,9 @@ func TestProcess_StopImmediately(t *testing.T) {
 // Test that SIGKILL is sent when gracefulStopTimeout is reached and properly terminates
 // the upstream command
 func TestProcess_ForceStopWithKill(t *testing.T) {
+	if runtime.GOOS == "windows" {
+		t.Skip("skipping SIGTERM test on Windows ")
+	}
 
 	expectedMessage := "test_sigkill"
 	binaryPath := getSimpleResponderPath()
@@ -405,7 +408,6 @@ func TestProcess_ForceStopWithKill(t *testing.T) {
 		Cmd:           fmt.Sprintf("%s --port %d --respond %s --silent --ignore-sig-term", binaryPath, port, expectedMessage),
 		Proxy:         fmt.Sprintf("http://127.0.0.1:%d", port),
 		CheckEndpoint: "/health",
-		CmdStop:       "taskkill /f /t /pid ${PID}",
 	}
 
 	process := NewProcess("stop_immediate", 2, config, debugLogger, debugLogger)
@@ -465,3 +467,27 @@ func TestProcess_StopCmd(t *testing.T) {
 	process.StopImmediately()
 	assert.Equal(t, process.CurrentState(), StateStopped)
 }
+
+func TestProcess_EnvironmentSetCorrectly(t *testing.T) {
+	expectedMessage := "test_env_not_emptied"
+	config := getTestSimpleResponderConfig(expectedMessage)
+
+	// ensure that the the default config does not blank out the inherited environment
+	configWEnv := config
+
+	// ensure the additiona variables are appended to the process' environment
+	configWEnv.Env = append(configWEnv.Env, "TEST_ENV1=1", "TEST_ENV2=2")
+
+	process1 := NewProcess("env_test", 2, config, debugLogger, debugLogger)
+	process2 := NewProcess("env_test", 2, configWEnv, debugLogger, debugLogger)
+
+	process1.start()
+	defer process1.Stop()
+	process2.start()
+	defer process2.Stop()
+
+	assert.NotZero(t, len(process1.cmd.Environ()))
+	assert.NotZero(t, len(process2.cmd.Environ()))
+	assert.Equal(t, len(process1.cmd.Environ())+2, len(process2.cmd.Environ()), "process2 should have 2 more environment variables than process1")
+
+}

proxy/proxymanager.go

@@ -365,6 +365,21 @@ func (pm *ProxyManager) proxyOAIHandler(c *gin.Context) {
 		}
 	}
 
+	// issue #174 strip parameters from the JSON body
+	stripParams, err := pm.config.Models[realModelName].Filters.SanitizedStripParams()
+	if err != nil { // just log it and continue
+		pm.proxyLogger.Errorf("Error sanitizing strip params string: %s, %s", pm.config.Models[realModelName].Filters.StripParams, err.Error())
+	} else {
+		for _, param := range stripParams {
+			pm.proxyLogger.Debugf("<%s> stripping param: %s", realModelName, param)
+			bodyBytes, err = sjson.DeleteBytes(bodyBytes, param)
+			if err != nil {
+				pm.sendErrorResponse(c, http.StatusInternalServerError, fmt.Sprintf("error deleting parameter %s from request", param))
+				return
+			}
+		}
+	}
+
 	c.Request.Body = io.NopCloser(bytes.NewBuffer(bodyBytes))
 
 	// dechunk it as we already have all the body bytes see issue #11

proxy/proxymanager_test.go

@@ -623,3 +623,37 @@ func TestProxyManager_ChatContentLength(t *testing.T) {
 	assert.Equal(t, "81", response["h_content_length"])
 	assert.Equal(t, "model1", response["responseMessage"])
 }
+
+func TestProxyManager_FiltersStripParams(t *testing.T) {
+	modelConfig := getTestSimpleResponderConfig("model1")
+	modelConfig.Filters = ModelFilters{
+		StripParams: "temperature, model, stream",
+	}
+
+	config := AddDefaultGroupToConfig(Config{
+		HealthCheckTimeout: 15,
+		LogLevel:           "error",
+		Models: map[string]ModelConfig{
+			"model1": modelConfig,
+		},
+	})
+
+	proxy := New(config)
+	defer proxy.StopProcesses(StopWaitForInflightRequest)
+	reqBody := `{"model":"model1", "temperature":0.1, "x_param":"123", "y_param":"abc", "stream":true}`
+	req := httptest.NewRequest("POST", "/v1/chat/completions", bytes.NewBufferString(reqBody))
+	w := httptest.NewRecorder()
+
+	proxy.ServeHTTP(w, req)
+	assert.Equal(t, http.StatusOK, w.Code)
+	var response map[string]string
+	assert.NoError(t, json.Unmarshal(w.Body.Bytes(), &response))
+
+	// `temperature` and `stream` are gone but model remains
+	assert.Equal(t, `{"model":"model1", "x_param":"123", "y_param":"abc"}`, response["request_body"])
+
+	// assert.Nil(t, response["temperature"])
+	// assert.Equal(t, "123", response["x_param"])
+	// assert.Equal(t, "abc", response["y_param"])
+	// t.Logf("%v", response)
+}

ui/index.html

@@ -3,7 +3,11 @@
   <head>
     <meta charset="UTF-8" />
     <meta name="viewport" content="width=device-width, initial-scale=1.0" />
-    <link rel="icon" type="image/png" href="/favicon.ico" />
+    <link rel="icon" type="image/png" href="/favicon-96x96.png" sizes="96x96" />
+    <link rel="icon" type="image/svg+xml" href="/favicon.svg" />
+    <link rel="shortcut icon" href="/favicon.ico" />
+    <link rel="apple-touch-icon" sizes="180x180" href="/apple-touch-icon.png" />
+    <link rel="manifest" href="/site.webmanifest" />
     <title>llama-swap</title>
   </head>
   <body >

ui/public/site.webmanifest

@@ -0,0 +1,21 @@
+{
+  "name": "llama-swap",
+  "short_name": "llama-swap",
+  "icons": [
+    {
+      "src": "/web-app-manifest-192x192.png",
+      "sizes": "192x192",
+      "type": "image/png",
+      "purpose": "maskable"
+    },
+    {
+      "src": "/web-app-manifest-512x512.png",
+      "sizes": "512x512",
+      "type": "image/png",
+      "purpose": "maskable"
+    }
+  ],
+  "theme_color": "#ffffff",
+  "background_color": "#ffffff",
+  "display": "standalone"
+}

ui/src/App.tsx

@@ -10,10 +10,10 @@ function App() {
     <Router basename="/ui/">
       <APIProvider>
         <div>
-          <nav className="bg-surface border-b border-border p-4">
-            <div className="flex items-center justify-between mx-auto px-4">
-              <h1>llama-swap</h1>
-              <div className="flex space-x-4">
+          <nav className="bg-surface border-b border-border p-2 h-[75px]">
+            <div className="flex items-center justify-between mx-auto px-4 h-full">
+              <h1 className="flex items-center p-0">llama-swap</h1>
+              <div className="flex items-center space-x-4">
                 <NavLink to="/" className={({ isActive }) => (isActive ? "navlink active" : "navlink")}>
                   Logs
                 </NavLink>

ui/src/contexts/APIProvider.tsx

@@ -12,6 +12,7 @@ interface APIProviderType {
   models: Model[];
   listModels: () => Promise<Model[]>;
   unloadAllModels: () => Promise<void>;
+  loadModel: (model: string) => Promise<void>;
   enableProxyLogs: (enabled: boolean) => void;
   enableUpstreamLogs: (enabled: boolean) => void;
   enableModelUpdates: (enabled: boolean) => void;
@@ -57,9 +58,27 @@ export function APIProvider({ children }: APIProviderProps) {
   const enableProxyLogs = useCallback(
     (enabled: boolean) => {
       if (enabled) {
-        const eventSource = new EventSource("/logs/streamSSE/proxy");
-        eventSource.onmessage = handleProxyMessage;
-        proxyEventSource.current = eventSource;
+        let retryCount = 0;
+        const maxRetries = 3;
+        const initialDelay = 1000; // 1 second
+
+        const connect = () => {
+          const eventSource = new EventSource("/logs/streamSSE/proxy");
+
+          eventSource.onmessage = handleProxyMessage;
+          eventSource.onerror = () => {
+            eventSource.close();
+            if (retryCount < maxRetries) {
+              retryCount++;
+              const delay = initialDelay * Math.pow(2, retryCount - 1);
+              setTimeout(connect, delay);
+            }
+          };
+
+          proxyEventSource.current = eventSource;
+        };
+
+        connect();
       } else {
         proxyEventSource.current?.close();
         proxyEventSource.current = null;
@@ -71,15 +90,33 @@ export function APIProvider({ children }: APIProviderProps) {
   const enableUpstreamLogs = useCallback(
     (enabled: boolean) => {
       if (enabled) {
-        const eventSource = new EventSource("/logs/streamSSE/upstream");
-        eventSource.onmessage = handleUpstreamMessage;
-        upstreamEventSource.current = eventSource;
+        let retryCount = 0;
+        const maxRetries = 3;
+        const initialDelay = 1000; // 1 second
+
+        const connect = () => {
+          const eventSource = new EventSource("/logs/streamSSE/upstream");
+
+          eventSource.onmessage = handleUpstreamMessage;
+          eventSource.onerror = () => {
+            eventSource.close();
+            if (retryCount < maxRetries) {
+              retryCount++;
+              const delay = initialDelay * Math.pow(2, retryCount - 1);
+              setTimeout(connect, delay);
+            }
+          };
+
+          upstreamEventSource.current = eventSource;
+        };
+
+        connect();
       } else {
         upstreamEventSource.current?.close();
         upstreamEventSource.current = null;
       }
     },
-    [upstreamEventSource, handleUpstreamMessage]
+    [handleUpstreamMessage]
   );
 
   const enableModelUpdates = useCallback(
@@ -139,11 +176,26 @@ export function APIProvider({ children }: APIProviderProps) {
     }
   }, []);
 
+  const loadModel = useCallback(async (model: string) => {
+    try {
+      const response = await fetch(`/upstream/${model}/`, {
+        method: "GET",
+      });
+      if (!response.ok) {
+        throw new Error(`Failed to load model: ${response.status}`);
+      }
+    } catch (error) {
+      console.error("Failed to load model:", error);
+      throw error; // Re-throw to let calling code handle it
+    }
+  }, []);
+
   const value = useMemo(
     () => ({
       models,
       listModels,
       unloadAllModels,
+      loadModel,
       enableProxyLogs,
       enableUpstreamLogs,
       enableModelUpdates,
@@ -154,6 +206,7 @@ export function APIProvider({ children }: APIProviderProps) {
       models,
       listModels,
       unloadAllModels,
+      loadModel,
       enableProxyLogs,
       enableUpstreamLogs,
       enableModelUpdates,

ui/src/index.css

@@ -143,6 +143,10 @@
     @apply bg-surface p-2 px-4 text-sm rounded-full border border-2 transition-colors duration-200 border-btn-border;
   }
 
+  .btn:hover {
+    cursor: pointer;
+  }
+
   .btn--sm {
     @apply px-2 py-0.5 text-xs;
   }

ui/src/lib/Utils.ts

@@ -0,0 +1,18 @@
+export function processEvalTimes(text: string) {
+  const lines = text.match(/^ *eval time.*$/gm) || [];
+
+  let totalTokens = 0;
+  let totalTime = 0;
+
+  lines.forEach((line) => {
+    const tokensMatch = line.match(/\/\s*(\d+)\s*tokens/);
+    const timeMatch = line.match(/=\s*(\d+\.\d+)\s*ms/);
+
+    if (tokensMatch) totalTokens += parseFloat(tokensMatch[1]);
+    if (timeMatch) totalTime += parseFloat(timeMatch[1]);
+  });
+
+  const avgTokensPerSecond = totalTime > 0 ? totalTokens / (totalTime / 1000) : 0;
+
+  return [lines.length, totalTokens, Math.round(avgTokensPerSecond * 100) / 100];
+}

ui/src/pages/LogViewer.tsx

@@ -15,7 +15,7 @@ const LogViewer = () => {
   }, []);
 
   return (
-    <div className="flex flex-col gap-5">
+    <div className="flex flex-col gap-5" style={{ height: "calc(100vh - 125px)" }}>
       <LogPanel id="proxy" title="Proxy Logs" logData={proxyLogs} />
       <LogPanel id="upstream" title="Upstream Logs" logData={upstreamLogs} />
     </div>
@@ -30,11 +30,8 @@ interface LogPanelProps {
 }
 
 export const LogPanel = ({ id, title, logData, className }: LogPanelProps) => {
+  const [isCollapsed, setIsCollapsed] = usePersistentState(`logPanel-${id}-isCollapsed`, false);
   const [filterRegex, setFilterRegex] = useState("");
-  const [panelState, setPanelState] = usePersistentState<"hide" | "small" | "max">(
-    `logPanel-${id}-panelState`,
-    "small"
-  );
   const [fontSize, setFontSize] = usePersistentState<"xxs" | "xs" | "small" | "normal">(
     `logPanel-${id}-fontSize`,
     "normal"
@@ -60,14 +57,6 @@ export const LogPanel = ({ id, title, logData, className }: LogPanelProps) => {
     });
   }, []);
 
-  const togglePanelState = useCallback(() => {
-    setPanelState((prev) => {
-      if (prev === "small") return "max";
-      if (prev === "hide") return "small";
-      return "hide";
-    });
-  }, []);
-
   const fontSizeClass = useMemo(() => {
     switch (fontSize) {
       case "xxs":
@@ -101,20 +90,21 @@ export const LogPanel = ({ id, title, logData, className }: LogPanelProps) => {
   }, [filteredLogs]);
 
   return (
-    <div className={`bg-surface border border-border rounded-lg overflow-hidden flex flex-col ${className || ""}`}>
+    <div
+      className={`bg-surface border border-border rounded-lg overflow-hidden flex flex-col ${
+        !isCollapsed && "h-full"
+      } ${className || ""}`}
+    >
       <div className="p-4 border-b border-border bg-secondary">
         <div className="flex flex-col md:flex-row md:items-center md:justify-between gap-4">
           {/* Title - Always full width on mobile, normal on desktop */}
-          <div className="w-full md:w-auto" onClick={togglePanelState}>
+          <div className="w-full md:w-auto" onClick={() => setIsCollapsed(!isCollapsed)}>
             <h3 className="m-0 text-lg">{title}</h3>
           </div>
 
           <div className="flex flex-col sm:flex-row gap-4 w-full md:w-auto">
             {/* Sizing Buttons - Stacks vertically on mobile */}
             <div className="flex flex-wrap gap-2">
-              <button className="btn" onClick={togglePanelState}>
-                size: {panelState}
-              </button>
               <button className="btn" onClick={toggleFontSize}>
                 font: {fontSize}
               </button>
@@ -140,14 +130,11 @@ export const LogPanel = ({ id, title, logData, className }: LogPanelProps) => {
         </div>
       </div>
 
-      {panelState !== "hide" && (
-        <div className="flex-1 bg-background font-mono text-sm leading-[1.4] p-3">
+      {!isCollapsed && (
+        <div className="flex-1 bg-background font-mono text-sm p-3 overflow-hidden">
           <pre
             ref={preTagRef}
-            className={`flex-1 p-4 overflow-y-auto whitespace-pre min-h-0 ${textWrapClass} ${fontSizeClass}`}
-            style={{
-              maxHeight: panelState === "max" ? "1500px" : "500px",
-            }}
+            className={`h-full p-4 overflow-y-auto whitespace-pre min-h-0 ${textWrapClass} ${fontSizeClass}`}
           >
             {filteredLogs}
           </pre>
@@ -156,5 +143,4 @@ export const LogPanel = ({ id, title, logData, className }: LogPanelProps) => {
     </div>
   );
 };
-
 export default LogViewer;

ui/src/pages/Models.tsx

@@ -1,9 +1,10 @@
-import { useState, useEffect, useCallback } from "react";
+import { useState, useEffect, useCallback, useMemo } from "react";
 import { useAPI } from "../contexts/APIProvider";
 import { LogPanel } from "./LogViewer";
+import { processEvalTimes } from "../lib/Utils";
 
 export default function ModelsPage() {
-  const { models, enableModelUpdates, unloadAllModels, upstreamLogs, enableUpstreamLogs } = useAPI();
+  const { models, enableModelUpdates, unloadAllModels, loadModel, upstreamLogs, enableUpstreamLogs } = useAPI();
   const [isUnloading, setIsUnloading] = useState(false);
 
   useEffect(() => {
@@ -29,8 +30,12 @@ export default function ModelsPage() {
     }
   }, []);
 
+  const [totalLines, totalTokens, avgTokensPerSecond] = useMemo(() => {
+    return processEvalTimes(upstreamLogs);
+  }, [upstreamLogs]);
+
   return (
-    <div className="h-screen">
+    <div>
       <div className="flex flex-col md:flex-row gap-4">
         {/* Left Column */}
         <div className="w-full md:w-1/2 flex items-top">
@@ -43,6 +48,7 @@ export default function ModelsPage() {
               <thead>
                 <tr className="border-b border-primary">
                   <th className="text-left p-2">Name</th>
+                  <th className="text-left p-2"></th>
                   <th className="text-left p-2">State</th>
                 </tr>
               </thead>
@@ -50,10 +56,19 @@ export default function ModelsPage() {
                 {models.map((model) => (
                   <tr key={model.id} className="border-b hover:bg-secondary-hover border-border">
                     <td className="p-2">
-                      <a href={`/upstream/${model.id}/`} className="underline" target="top">
+                      <a href={`/upstream/${model.id}/`} className="underline" target="_blank">
                         {model.id}
                       </a>
                     </td>
+                    <td className="p-2">
+                      <button
+                        className="btn btn--sm"
+                        disabled={model.state !== "stopped"}
+                        onClick={() => loadModel(model.id)}
+                      >
+                        Load
+                      </button>
+                    </td>
                     <td className="p-2">
                       <span className={`status status--${model.state}`}>{model.state}</span>
                     </td>
@@ -65,8 +80,29 @@ export default function ModelsPage() {
         </div>
 
         {/* Right Column */}
-        <div className="w-full md:w-1/2  flex items-top">
-          <LogPanel id="modelsupstream" title="Upstream Logs" logData={upstreamLogs} className="h-full" />
+        <div className="w-full md:w-1/2 flex flex-col" style={{ height: "calc(100vh - 125px)" }}>
+          <div className="card mb-4 min-h-[250px]">
+            <h2>Log Stats</h2>
+            <p className="italic my-2">note: eval logs from llama-server</p>
+            <table className="w-full border border-gray-200">
+              <tbody>
+                <tr className="border-b border-gray-200">
+                  <td className="py-2 px-4 font-medium border-r border-gray-200">Requests</td>
+                  <td className="py-2 px-4 text-right">{totalLines}</td>
+                </tr>
+                <tr className="border-b border-gray-200">
+                  <td className="py-2 px-4 font-medium border-r border-gray-200">Total Tokens Generated</td>
+                  <td className="py-2 px-4 text-right">{totalTokens}</td>
+                </tr>
+                <tr>
+                  <td className="py-2 px-4 font-medium border-r border-gray-200">Average Tokens/Second</td>
+                  <td className="py-2 px-4 text-right">{avgTokensPerSecond}</td>
+                </tr>
+              </tbody>
+            </table>
+          </div>
+
+          <LogPanel id="modelsupstream" title="Upstream Logs" logData={upstreamLogs} />
         </div>
       </div>
     </div>