mirror of
https://github.com/R0m1k3/Loki.git
synced 2026-10-11 17:26:57 +02:00
gofmt -w sur les 6 fichiers non conformes (alignements de champs/maps uniquement)
This commit is contained in:
nathaninline committed
1 parent
71e7a5c369
commit
392f85ec3b
6 files changed
+43
-43
No files matched your search
@@ -8,12 +8,12 @@ func TestCompiledFile(t *testing.T) {
|
||||
` ggml-threading.cpp`: "ggml-threading.cpp",
|
||||
` ggml-quants.c`: "ggml-quants.c",
|
||||
// Make (Linux) et Ninja.
|
||||
`[ 45%] Building CXX object src/CMakeFiles/llama.dir/llama.cpp.o`: "llama.cpp",
|
||||
`[ 45%] Building CXX object src/CMakeFiles/llama.dir/llama.cpp.o`: "llama.cpp",
|
||||
`[123/456] Building CUDA object ggml/src/ggml-cuda/CMakeFiles/ggml-cuda.dir/acc.cu.o`: "acc.cu",
|
||||
// La ligne de commande nvcc géante ne doit PAS être prise pour un fichier.
|
||||
` C:\...\nvcc.exe -x cu ... -o ggml-cuda.dir\Release\acc.obj "C:\...\acc.cu"`: "",
|
||||
`Building Custom Rule C:/ProgramData/jean/...`: "",
|
||||
`-- UI: running npm install`: "",
|
||||
`Building Custom Rule C:/ProgramData/jean/...`: "",
|
||||
`-- UI: running npm install`: "",
|
||||
}
|
||||
for in, want := range cases {
|
||||
if got := compiledFile(in); got != want {
|
||||
|
||||
+19
-19
@@ -14,13 +14,13 @@ import (
|
||||
|
||||
// benchResult captures the timings llama.cpp returns from /completion.
|
||||
type benchResult struct {
|
||||
PromptN int `json:"prompt_n"`
|
||||
PromptMs float64 `json:"prompt_ms"`
|
||||
PromptPerSecond float64 `json:"prompt_per_second"`
|
||||
PredictedN int `json:"predicted_n"`
|
||||
PredictedMs float64 `json:"predicted_ms"`
|
||||
PredictedPerSec float64 `json:"predicted_per_second"`
|
||||
Elapsed float64 `json:"elapsed_sec"`
|
||||
PromptN int `json:"prompt_n"`
|
||||
PromptMs float64 `json:"prompt_ms"`
|
||||
PromptPerSecond float64 `json:"prompt_per_second"`
|
||||
PredictedN int `json:"predicted_n"`
|
||||
PredictedMs float64 `json:"predicted_ms"`
|
||||
PredictedPerSec float64 `json:"predicted_per_second"`
|
||||
Elapsed float64 `json:"elapsed_sec"`
|
||||
}
|
||||
|
||||
// benchCorpus is a varied passage used to defeat speculative decoding
|
||||
@@ -71,12 +71,12 @@ func runBench(nPrompt, nPredict int) (*benchResult, error) {
|
||||
// Use the same endpoint your real chat hits, so the comparison is honest
|
||||
// (chat template, reasoning, OpenAI-compat layer all included).
|
||||
payload := map[string]any{
|
||||
"model": "jean",
|
||||
"messages": []Message{{Role: "user", Content: prompt + "\n\nContinue this passage with another 1000+ words of original varied prose, mixing French and English narrative paragraphs on different topics."}},
|
||||
"max_tokens": nPredict,
|
||||
"stream": false,
|
||||
"temperature": 0.7,
|
||||
"cache_prompt": false,
|
||||
"model": "jean",
|
||||
"messages": []Message{{Role: "user", Content: prompt + "\n\nContinue this passage with another 1000+ words of original varied prose, mixing French and English narrative paragraphs on different topics."}},
|
||||
"max_tokens": nPredict,
|
||||
"stream": false,
|
||||
"temperature": 0.7,
|
||||
"cache_prompt": false,
|
||||
}
|
||||
body, _ := json.Marshal(payload)
|
||||
url := fmt.Sprintf("http://localhost:%d/v1/chat/completions", port)
|
||||
@@ -92,12 +92,12 @@ func runBench(nPrompt, nPredict int) (*benchResult, error) {
|
||||
defer resp.Body.Close()
|
||||
var parsed struct {
|
||||
Timings struct {
|
||||
PromptN int `json:"prompt_n"`
|
||||
PromptMs float64 `json:"prompt_ms"`
|
||||
PromptPerSecond float64 `json:"prompt_per_second"`
|
||||
PredictedN int `json:"predicted_n"`
|
||||
PredictedMs float64 `json:"predicted_ms"`
|
||||
PredictedPerSec float64 `json:"predicted_per_second"`
|
||||
PromptN int `json:"prompt_n"`
|
||||
PromptMs float64 `json:"prompt_ms"`
|
||||
PromptPerSecond float64 `json:"prompt_per_second"`
|
||||
PredictedN int `json:"predicted_n"`
|
||||
PredictedMs float64 `json:"predicted_ms"`
|
||||
PredictedPerSec float64 `json:"predicted_per_second"`
|
||||
} `json:"timings"`
|
||||
Usage struct {
|
||||
PromptTokens int `json:"prompt_tokens"`
|
||||
|
||||
@@ -466,4 +466,3 @@ func demuxTunnelStream(stream net.Conn, httpLn, oaiLn *chanListener) {
|
||||
}
|
||||
httpLn.push(pc)
|
||||
}
|
||||
|
||||
@@ -17,9 +17,9 @@ import (
|
||||
|
||||
func relaunchDetachedApp() {
|
||||
const (
|
||||
detachedProcess = 0x00000008
|
||||
createNoWindow = 0x08000000
|
||||
createNewProcGrp = 0x00000200
|
||||
detachedProcess = 0x00000008
|
||||
createNoWindow = 0x08000000
|
||||
createNewProcGrp = 0x00000200
|
||||
)
|
||||
exe, err := os.Executable()
|
||||
if err != nil {
|
||||
|
||||
@@ -105,17 +105,17 @@ func trayIcon() []byte {
|
||||
|
||||
// Conteneur ICO : ICONDIR (6) + 1 ICONDIRENTRY (16) + PNG.
|
||||
var ico bytes.Buffer
|
||||
binary.Write(&ico, binary.LittleEndian, uint16(0)) // réservé
|
||||
binary.Write(&ico, binary.LittleEndian, uint16(1)) // type = icône
|
||||
binary.Write(&ico, binary.LittleEndian, uint16(1)) // nombre d'images
|
||||
ico.WriteByte(n) // largeur
|
||||
ico.WriteByte(n) // hauteur
|
||||
ico.WriteByte(0) // couleurs
|
||||
ico.WriteByte(0) // réservé
|
||||
binary.Write(&ico, binary.LittleEndian, uint16(1)) // plans
|
||||
binary.Write(&ico, binary.LittleEndian, uint16(32)) // bits/pixel
|
||||
binary.Write(&ico, binary.LittleEndian, uint32(len(p))) // taille données
|
||||
binary.Write(&ico, binary.LittleEndian, uint32(6+16)) // offset données
|
||||
binary.Write(&ico, binary.LittleEndian, uint16(0)) // réservé
|
||||
binary.Write(&ico, binary.LittleEndian, uint16(1)) // type = icône
|
||||
binary.Write(&ico, binary.LittleEndian, uint16(1)) // nombre d'images
|
||||
ico.WriteByte(n) // largeur
|
||||
ico.WriteByte(n) // hauteur
|
||||
ico.WriteByte(0) // couleurs
|
||||
ico.WriteByte(0) // réservé
|
||||
binary.Write(&ico, binary.LittleEndian, uint16(1)) // plans
|
||||
binary.Write(&ico, binary.LittleEndian, uint16(32)) // bits/pixel
|
||||
binary.Write(&ico, binary.LittleEndian, uint32(len(p))) // taille données
|
||||
binary.Write(&ico, binary.LittleEndian, uint32(6+16)) // offset données
|
||||
ico.Write(p)
|
||||
return ico.Bytes()
|
||||
}
|
||||
@@ -116,12 +116,12 @@ func newWebMux() *http.ServeMux {
|
||||
api("/api/restart", svcHandler("restart"))
|
||||
api("/api/bench", handleBench)
|
||||
api("/api/bench/last", handleBenchLast)
|
||||
api("/api/chat", handleChat) // flux d'ABONNEMENT (SSE) : rejoue + suit le fil
|
||||
api("/api/chat/send", handleChatSend) // envoie un message (lance la génération détachée)
|
||||
api("/api/chat/stop", handleChatStop) // interrompt la génération en cours
|
||||
api("/api/chat/reset", handleChatReset) // nouvelle conversation (pour tous les appareils)
|
||||
api("/api/chat/state", handleChatState) // instantané léger {seq, generating, ctx_used}
|
||||
api("/api/e2e/chat", handleE2EChat) // même flux mais chiffré E2E (boîte noire via le relais)
|
||||
api("/api/chat", handleChat) // flux d'ABONNEMENT (SSE) : rejoue + suit le fil
|
||||
api("/api/chat/send", handleChatSend) // envoie un message (lance la génération détachée)
|
||||
api("/api/chat/stop", handleChatStop) // interrompt la génération en cours
|
||||
api("/api/chat/reset", handleChatReset) // nouvelle conversation (pour tous les appareils)
|
||||
api("/api/chat/state", handleChatState) // instantané léger {seq, generating, ctx_used}
|
||||
api("/api/e2e/chat", handleE2EChat) // même flux mais chiffré E2E (boîte noire via le relais)
|
||||
return mux
|
||||
}
|
||||
|
||||
@@ -537,6 +537,7 @@ func handleAgentToggle(w http.ResponseWriter, r *http.Request) {
|
||||
//
|
||||
// GET → {enabled, url, reachable}
|
||||
// POST {enabled, url} → enregistre CRAWL4AI_URL + le drapeau .internet_enabled
|
||||
//
|
||||
// handleAPIKey expose et pilote la clé d'accès à l'endpoint compatible OpenAI
|
||||
// (llama-server /v1). GET renvoie l'état ; POST {action:"generate"|"set"|"clear",
|
||||
// key?} l'écrit puis redémarre le service (llama-server lit --api-key au lancement).
|
||||
|
||||
Reference in new issue
Block a user