refactor: 训练日志组件化与 Mock 数据增强
拆分 TrainingTaskOverview 组件与 trainingLogModel 状态模型，TrainingLogView 大幅瘦身；Mock 新增按文件路由的训练日志内容与更真实的 GPU 进程占用数据，adapter 类型收敛为 AxiosAdapter，配套新增 mock 内容回归脚本。
This commit is contained in:
caoxiaozhu
@@ -3,7 +3,7 @@
|
||||
* 拦截所有 API 请求并返回 mock 数据
|
||||
* 通过 URL + method 路由到对应的 mock 响应
|
||||
*/
|
||||
import type { AxiosInstance, AxiosRequestConfig } from 'axios'
|
||||
import type { AxiosAdapter, AxiosInstance, AxiosRequestConfig } from 'axios'
|
||||
import {
|
||||
mockLoginOk,
|
||||
mockHealth,
|
||||
@@ -21,6 +21,7 @@ import {
|
||||
mockLogFiles,
|
||||
mockTrainingLogFiles,
|
||||
mockLogContent,
|
||||
mockTrainingLogContents,
|
||||
} from './data'
|
||||
import {
|
||||
activateDatasetVersion,
|
||||
@@ -164,7 +165,8 @@ async function handleMock(config: AxiosRequestConfig) {
|
||||
}
|
||||
m = url.match(/^\/dataset-manage\/([^/]+)$/)
|
||||
if (m && method === 'get') {
|
||||
const found = mockDatasets.find((x) => String(x.id) === m[1])
|
||||
const datasetId = m[1]
|
||||
const found = mockDatasets.find((x) => String(x.id) === datasetId)
|
||||
return found ? ok(found) : fail('数据集不存在', 404)
|
||||
}
|
||||
if (m && (method === 'put' || method === 'delete')) {
|
||||
@@ -261,7 +263,8 @@ async function handleMock(config: AxiosRequestConfig) {
|
||||
}
|
||||
m = url.match(/^\/fine-tune\/progress\/([^/]+)$/)
|
||||
if (m && method === 'get') {
|
||||
const task = mockFineTuneList.find((t) => String(t.id) === m[1])
|
||||
const taskId = m[1]
|
||||
const task = mockFineTuneList.find((t) => String(t.id) === taskId)
|
||||
if (!task) return fail('任务不存在', 404)
|
||||
if (task.status === 'running') {
|
||||
return ok({
|
||||
@@ -276,7 +279,8 @@ async function handleMock(config: AxiosRequestConfig) {
|
||||
}
|
||||
m = url.match(/^\/fine-tune\/([^/]+)$/)
|
||||
if (m && method === 'get') {
|
||||
const found = mockFineTuneList.find((x) => String(x.id) === m[1])
|
||||
const taskId = m[1]
|
||||
const found = mockFineTuneList.find((x) => String(x.id) === taskId)
|
||||
return found ? ok(found) : fail('任务不存在', 404)
|
||||
}
|
||||
m = url.match(/^\/fine-tune\/stop\/([^/]+)$/)
|
||||
@@ -296,7 +300,8 @@ async function handleMock(config: AxiosRequestConfig) {
|
||||
}
|
||||
m = url.match(/^\/model-compare\/([^/]+)$/)
|
||||
if (m && method === 'get') {
|
||||
const found = mockCompareList.find((x) => String(x.id) === m[1])
|
||||
const compareId = m[1]
|
||||
const found = mockCompareList.find((x) => String(x.id) === compareId)
|
||||
return found ? ok(found) : fail('任务不存在', 404)
|
||||
}
|
||||
if (m && method === 'delete') return ok({ deleted: m[1] })
|
||||
@@ -363,7 +368,14 @@ async function handleMock(config: AxiosRequestConfig) {
|
||||
if (url === '/log-files' && method === 'get') return ok(mockLogFiles)
|
||||
if (url === '/log-content' && method === 'get') return ok(mockLogContent)
|
||||
if (url === '/training-log-files' && method === 'get') return ok(mockTrainingLogFiles)
|
||||
if (url === '/training-log-content' && method === 'get') return ok(mockLogContent)
|
||||
if (url === '/training-log-content' && method === 'get') {
|
||||
const file = String(params.file || '')
|
||||
return ok(mockTrainingLogContents[file] || {
|
||||
file,
|
||||
size: '0 KB',
|
||||
content: `[Mock] 未找到训练日志内容：${file}`,
|
||||
})
|
||||
}
|
||||
|
||||
// 未匹配的请求 → 兜底返回空成功（避免阻断 UI）
|
||||
console.warn('[Mock] 未匹配路由:', method.toUpperCase(), url, params)
|
||||
@@ -380,12 +392,13 @@ function safeJSON(str: string) {
|
||||
|
||||
/** 给 axios instance 安装 mock adapter */
|
||||
export function installMockAdapter(instance: AxiosInstance) {
|
||||
instance.defaults.adapter = async (config: AxiosRequestConfig) => {
|
||||
const adapter = async (config: AxiosRequestConfig) => {
|
||||
try {
|
||||
const response = await handleMock(config)
|
||||
return response
|
||||
} catch (e: any) {
|
||||
return fail(e.message || 'Mock 错误', 500, config)
|
||||
} catch (error: unknown) {
|
||||
return fail(error instanceof Error ? error.message : 'Mock 错误', 500, config)
|
||||
}
|
||||
}
|
||||
instance.defaults.adapter = adapter as AxiosAdapter
|
||||
}
|
||||
|
||||
@@ -58,37 +58,40 @@ export const mockSystemInfo: SystemInfo = {
|
||||
id: 0,
|
||||
uuid: 'GPU-MOCK-A800-00',
|
||||
name: 'NVIDIA A800',
|
||||
status: 'idle',
|
||||
gpu_percent: 0,
|
||||
memory_used_gb: 0,
|
||||
status: 'busy',
|
||||
gpu_percent: 74,
|
||||
memory_used_gb: 41.8,
|
||||
memory_total_gb: 80,
|
||||
memory_percent: 0,
|
||||
temperature: 32,
|
||||
power_w: 38,
|
||||
memory_percent: 52.3,
|
||||
temperature: 63,
|
||||
power_w: 286,
|
||||
power_limit_w: 400,
|
||||
fan_speed: 0,
|
||||
fan_speed: 51,
|
||||
clock_mhz: 1410,
|
||||
driver_version: '535.86.10',
|
||||
processes: [],
|
||||
processes: [
|
||||
{ pid: 28741, name: 'python', task_name: 'medical-cpt-001 / rank 0', user: 'trainer', memory_used_gb: 39.6 },
|
||||
{ pid: 28768, name: 'python', task_name: '训练指标采集', user: 'trainer', memory_used_gb: 2.2 },
|
||||
],
|
||||
},
|
||||
{
|
||||
id: 1,
|
||||
uuid: 'GPU-MOCK-A800-01',
|
||||
name: 'NVIDIA A800',
|
||||
status: 'busy',
|
||||
gpu_percent: 28,
|
||||
memory_used_gb: 22.5,
|
||||
gpu_percent: 71,
|
||||
memory_used_gb: 42.1,
|
||||
memory_total_gb: 80,
|
||||
memory_percent: 28.1,
|
||||
temperature: 52,
|
||||
power_w: 165,
|
||||
memory_percent: 52.6,
|
||||
temperature: 62,
|
||||
power_w: 279,
|
||||
power_limit_w: 400,
|
||||
fan_speed: 32,
|
||||
fan_speed: 49,
|
||||
clock_mhz: 1410,
|
||||
driver_version: '535.86.10',
|
||||
processes: [
|
||||
{ pid: 18421, name: 'python', task_name: '指令微调任务', user: 'trainer', memory_used_gb: 18.6 },
|
||||
{ pid: 18503, name: 'python', task_name: '训练指标采集', user: 'trainer', memory_used_gb: 3.9 },
|
||||
{ pid: 28742, name: 'python', task_name: 'medical-cpt-001 / rank 1', user: 'trainer', memory_used_gb: 39.9 },
|
||||
{ pid: 28769, name: 'python', task_name: '训练指标采集', user: 'trainer', memory_used_gb: 2.2 },
|
||||
],
|
||||
},
|
||||
{
|
||||
@@ -249,7 +252,7 @@ export const mockLocalModels = {
|
||||
}
|
||||
|
||||
// ============ 数据集 ============
|
||||
export const mockDatasets: DatasetItem[] = [
|
||||
export const mockDatasets: DatasetItem[] = ([
|
||||
{
|
||||
id: 1,
|
||||
name: '金融问答-训练集',
|
||||
@@ -273,7 +276,7 @@ export const mockDatasets: DatasetItem[] = [
|
||||
{ id: 8, name: '通用指令构造集', type: 'train', storage_type: 'local', source: 'task', task_id: 492015, size: '148 MB', count: 12600, description: '由指令微调数据构造任务生成', create_time: '2026-07-09T01:42:00Z' },
|
||||
{ id: 9, name: '用户反馈脱敏集', type: 'test', storage_type: 'minio', source: 'task', task_id: 731948, size: '72 MB', count: 9340, description: '由敏感信息脱敏任务生成', create_time: '2026-07-09T09:18:00Z' },
|
||||
{ id: 10, name: '多轮对话增强集', type: 'eval', storage_type: 'local', source: 'task', task_id: 582012, size: '41 MB', count: 2780, description: '由多轮对话拼接任务生成', create_time: '2026-07-10T02:06:00Z' },
|
||||
].map((dataset) => ({
|
||||
] satisfies DatasetItem[]).map((dataset) => ({
|
||||
...dataset,
|
||||
files: dataset.files?.length
|
||||
? dataset.files
|
||||
@@ -339,12 +342,181 @@ export const mockDatasetPreviews: Record<string, string> = {
|
||||
|
||||
// ============ 训练任务 ============
|
||||
export const mockFineTuneList: FineTuneTask[] = [
|
||||
{ id: 103942, name: 'finance-sft-001', description: '金融领域 SFT 训练', status: 'completed', train_type: 'SFT', train_method: 'lora', template: 'qwen', base_model: 1, train_dataset_id: 1, gpus: [0], progress: 100, train_duration: '2小时18分钟', create_time: '2026-01-15T08:00:00Z' },
|
||||
{ id: 349102, name: 'legal-sft-002', description: '法律文书 SFT', status: 'completed', train_type: 'SFT', train_method: 'lora', template: 'qwen', base_model: 1, train_dataset_id: 2, gpus: [1], progress: 100, train_duration: '1小时46分钟', create_time: '2026-01-18T10:00:00Z' },
|
||||
{ id: 849301, name: 'medical-cpt-001', description: '医疗领域继续预训练', status: 'running', train_type: 'CPT', train_method: 'lora', template: 'qwen2_5', base_model: 2, train_dataset_id: 6, gpus: [0, 1], progress: 64, train_duration: '36分钟', create_time: '2026-02-05T09:00:00Z' },
|
||||
{ id: 593021, name: 'service-dpo-001', description: '客服对话偏好训练', status: 'pending', train_type: 'DPO', train_method: 'lora', template: 'qwen', base_model: 1, train_dataset_id: 3, gpus: [2], progress: 0, train_duration: '-', create_time: '2026-02-08T14:00:00Z' },
|
||||
{ id: 201948, name: 'finance-sft-002', description: '金融领域二轮微调', status: 'failed', train_type: 'SFT', train_method: 'lora', template: 'qwen', base_model: 1, train_dataset_id: 1, gpus: [3], progress: 32, train_duration: '18分钟', create_time: '2026-02-10T11:00:00Z' },
|
||||
{ id: 940212, name: 'general-sft-001', description: '通用能力微调', status: 'completed', train_type: 'SFT', train_method: 'full', template: 'llama3', base_model: 3, train_dataset_id: 3, gpus: [0, 2], progress: 100, train_duration: '3小时05分钟', create_time: '2026-02-12T13:00:00Z' },
|
||||
{
|
||||
id: 103942,
|
||||
name: 'finance-sft-001',
|
||||
description: '金融领域 SFT 训练',
|
||||
status: 'completed',
|
||||
train_type: 'SFT',
|
||||
train_method: 'lora',
|
||||
template: 'qwen',
|
||||
base_model: 1,
|
||||
train_dataset_id: 1,
|
||||
output_model_name: 'qwen2.5-7b-finance-sft-v1',
|
||||
auto_merge: true,
|
||||
gpus: [0],
|
||||
batch_size: 8,
|
||||
learning_rate: 0.00002,
|
||||
n_epochs: 3,
|
||||
save_steps: 100,
|
||||
lr_scheduler_type: 'cosine',
|
||||
max_length: 2048,
|
||||
warmup_ratio: 0.05,
|
||||
weight_decay: 0.01,
|
||||
lora_rank: 16,
|
||||
lora_alpha: 32,
|
||||
lora_dropout: 0.05,
|
||||
quantization_bit: 0,
|
||||
process_id: 12345,
|
||||
progress: 100,
|
||||
train_duration: '2小时18分钟',
|
||||
create_time: '2026-01-15T08:00:00Z',
|
||||
},
|
||||
{
|
||||
id: 349102,
|
||||
name: 'legal-sft-002',
|
||||
description: '法律文书 SFT',
|
||||
status: 'completed',
|
||||
train_type: 'SFT',
|
||||
train_method: 'lora',
|
||||
template: 'qwen',
|
||||
base_model: 1,
|
||||
train_dataset_id: 2,
|
||||
output_model_name: 'qwen2.5-7b-legal-sft-v2',
|
||||
auto_merge: true,
|
||||
gpus: [1],
|
||||
batch_size: 4,
|
||||
learning_rate: 0.000015,
|
||||
n_epochs: 4,
|
||||
save_steps: 120,
|
||||
lr_scheduler_type: 'linear',
|
||||
max_length: 4096,
|
||||
warmup_ratio: 0.03,
|
||||
weight_decay: 0.01,
|
||||
lora_rank: 32,
|
||||
lora_alpha: 64,
|
||||
lora_dropout: 0.05,
|
||||
quantization_bit: 0,
|
||||
process_id: 12350,
|
||||
progress: 100,
|
||||
train_duration: '1小时46分钟',
|
||||
create_time: '2026-01-18T10:00:00Z',
|
||||
},
|
||||
{
|
||||
id: 849301,
|
||||
name: 'medical-cpt-001',
|
||||
description: '医疗领域继续预训练',
|
||||
status: 'running',
|
||||
train_type: 'CPT',
|
||||
train_method: 'lora',
|
||||
template: 'qwen2_5',
|
||||
base_model: 2,
|
||||
train_dataset_id: 6,
|
||||
output_model_name: 'qwen2.5-14b-medical-cpt-v1',
|
||||
auto_merge: true,
|
||||
gpus: [0, 1],
|
||||
batch_size: 2,
|
||||
learning_rate: 0.0001,
|
||||
n_epochs: 3,
|
||||
save_steps: 200,
|
||||
lr_scheduler_type: 'cosine',
|
||||
max_length: 4096,
|
||||
warmup_ratio: 0.03,
|
||||
weight_decay: 0.01,
|
||||
lora_rank: 16,
|
||||
lora_alpha: 32,
|
||||
lora_dropout: 0.05,
|
||||
quantization_bit: 0,
|
||||
process_id: 28741,
|
||||
progress: 64,
|
||||
train_duration: '36分钟',
|
||||
create_time: '2026-07-13T06:40:00Z',
|
||||
},
|
||||
{
|
||||
id: 593021,
|
||||
name: 'service-dpo-001',
|
||||
description: '客服对话偏好训练',
|
||||
status: 'pending',
|
||||
train_type: 'DPO',
|
||||
train_method: 'lora',
|
||||
template: 'qwen',
|
||||
base_model: 1,
|
||||
train_dataset_id: 3,
|
||||
output_model_name: 'qwen2.5-7b-service-dpo-v1',
|
||||
auto_merge: true,
|
||||
gpus: [2],
|
||||
batch_size: 4,
|
||||
learning_rate: 0.000005,
|
||||
n_epochs: 2,
|
||||
save_steps: 100,
|
||||
lr_scheduler_type: 'cosine',
|
||||
max_length: 2048,
|
||||
warmup_ratio: 0.1,
|
||||
weight_decay: 0,
|
||||
lora_rank: 16,
|
||||
lora_alpha: 32,
|
||||
lora_dropout: 0.1,
|
||||
quantization_bit: 0,
|
||||
progress: 0,
|
||||
train_duration: '等待调度',
|
||||
create_time: '2026-02-08T14:00:00Z',
|
||||
},
|
||||
{
|
||||
id: 201948,
|
||||
name: 'finance-sft-002',
|
||||
description: '金融领域二轮微调',
|
||||
status: 'failed',
|
||||
train_type: 'SFT',
|
||||
train_method: 'lora',
|
||||
template: 'qwen',
|
||||
base_model: 1,
|
||||
train_dataset_id: 1,
|
||||
output_model_name: 'qwen2.5-7b-finance-sft-v2',
|
||||
auto_merge: false,
|
||||
gpus: [3],
|
||||
batch_size: 8,
|
||||
learning_rate: 0.00002,
|
||||
n_epochs: 3,
|
||||
save_steps: 100,
|
||||
lr_scheduler_type: 'cosine',
|
||||
max_length: 2048,
|
||||
warmup_ratio: 0.05,
|
||||
weight_decay: 0.01,
|
||||
lora_rank: 16,
|
||||
lora_alpha: 32,
|
||||
lora_dropout: 0.05,
|
||||
quantization_bit: 0,
|
||||
process_id: 27654,
|
||||
progress: 32,
|
||||
train_duration: '18分钟',
|
||||
create_time: '2026-02-10T11:00:00Z',
|
||||
},
|
||||
{
|
||||
id: 940212,
|
||||
name: 'general-sft-001',
|
||||
description: '通用能力全参数微调',
|
||||
status: 'completed',
|
||||
train_type: 'SFT',
|
||||
train_method: 'full',
|
||||
template: 'llama3',
|
||||
base_model: 3,
|
||||
train_dataset_id: 3,
|
||||
output_model_name: 'llama3-8b-general-sft-v1',
|
||||
auto_merge: false,
|
||||
gpus: [0, 2],
|
||||
batch_size: 2,
|
||||
learning_rate: 0.00001,
|
||||
n_epochs: 2,
|
||||
save_steps: 250,
|
||||
lr_scheduler_type: 'cosine',
|
||||
max_length: 4096,
|
||||
warmup_ratio: 0.03,
|
||||
weight_decay: 0.1,
|
||||
process_id: 26318,
|
||||
progress: 100,
|
||||
train_duration: '3小时05分钟',
|
||||
create_time: '2026-02-12T13:00:00Z',
|
||||
},
|
||||
]
|
||||
|
||||
// ============ 模型推理/对比 ============
|
||||
@@ -549,11 +721,48 @@ export const mockLogFiles: LogFile[] = [
|
||||
]
|
||||
|
||||
export const mockTrainingLogFiles: TrainingLogFile[] = [
|
||||
{ file: 'medical-cpt-001_pid28741.log', name: 'medical-cpt-001', size: '6.8 MB', pid: 28741, date: '2026-07-13' },
|
||||
{ file: 'qwen-ft-finance-001_pid12345.log', name: 'finance-sft-001', size: '4.5 MB', pid: 12345, date: '2026-02-15' },
|
||||
{ file: 'llama3-ft-customer-service_pid12346.log', name: 'service-dpo-001', size: '2.1 MB', pid: 12346, date: '2026-02-18' },
|
||||
{ file: 'qwen-ft-legal-002_pid12350.log', name: 'legal-sft-002', size: '5.8 MB', pid: 12350, date: '2026-02-20' },
|
||||
]
|
||||
|
||||
const medicalTrainingLogLines = [
|
||||
'[2026-07-13 14:40:01] INFO: Launching distributed training with torchrun --nproc_per_node=2',
|
||||
'[2026-07-13 14:40:02] INFO: Process rank: 0, world size: 2, device: cuda:0, distributed training: True',
|
||||
'[2026-07-13 14:40:04] INFO: Loading tokenizer from /data/models/qwen2.5-14b-instruct',
|
||||
'[2026-07-13 14:40:16] INFO: Loading checkpoint shards: 100% | 8/8 | 00:12',
|
||||
'[2026-07-13 14:40:18] INFO: Loading dataset 医疗问答-训练集 (9,800 samples)',
|
||||
'[2026-07-13 14:40:27] INFO: Tokenizing dataset: 100% | 9,800/9,800 | 00:09',
|
||||
'[2026-07-13 14:40:28] INFO: LoRA config: rank=16, alpha=32, dropout=0.05, target_modules=q_proj,k_proj,v_proj,o_proj',
|
||||
'[2026-07-13 14:40:29] INFO: Trainable params: 83,886,080 / 14,787,584,000 (0.5673%)',
|
||||
'[2026-07-13 14:40:30] INFO: ***** Running training *****',
|
||||
'[2026-07-13 14:40:30] INFO: Num examples = 9,800',
|
||||
'[2026-07-13 14:40:30] INFO: Num Epochs = 3',
|
||||
'[2026-07-13 14:40:30] INFO: Instantaneous batch size per device = 2',
|
||||
'[2026-07-13 14:40:30] INFO: Total train batch size = 32',
|
||||
'[2026-07-13 14:40:30] INFO: Gradient Accumulation steps = 8',
|
||||
'[2026-07-13 14:40:30] INFO: Total optimization steps = 921',
|
||||
'[2026-07-13 14:42:18] INFO: step=40 {\'loss\': 2.684, \'grad_norm\': 1.184, \'learning_rate\': 9.82e-05, \'epoch\': 0.13}',
|
||||
'[2026-07-13 14:44:26] INFO: step=80 {\'loss\': 2.312, \'grad_norm\': 1.092, \'learning_rate\': 9.68e-05, \'epoch\': 0.26}',
|
||||
'[2026-07-13 14:46:34] INFO: step=120 {\'loss\': 2.084, \'grad_norm\': 1.037, \'learning_rate\': 9.43e-05, \'epoch\': 0.39}',
|
||||
'[2026-07-13 14:48:42] INFO: step=160 {\'loss\': 1.932, \'grad_norm\': 0.986, \'learning_rate\': 9.08e-05, \'epoch\': 0.52}',
|
||||
'[2026-07-13 14:50:49] INFO: step=200 {\'loss\': 1.801, \'grad_norm\': 0.944, \'learning_rate\': 8.64e-05, \'epoch\': 0.65}',
|
||||
'[2026-07-13 14:50:54] INFO: Saving checkpoint to /data/checkpoints/medical-cpt-001/checkpoint-200',
|
||||
'[2026-07-13 14:52:58] INFO: step=240 {\'loss\': 1.696, \'grad_norm\': 0.913, \'learning_rate\': 8.15e-05, \'epoch\': 0.78}',
|
||||
'[2026-07-13 14:55:06] INFO: step=280 {\'loss\': 1.611, \'grad_norm\': 0.887, \'learning_rate\': 7.61e-05, \'epoch\': 0.91}',
|
||||
'[2026-07-13 14:57:13] INFO: step=320 {\'loss\': 1.532, \'grad_norm\': 0.852, \'learning_rate\': 7.06e-05, \'epoch\': 1.04}',
|
||||
'[2026-07-13 14:59:21] INFO: step=360 {\'loss\': 1.461, \'grad_norm\': 0.829, \'learning_rate\': 6.49e-05, \'epoch\': 1.17}',
|
||||
'[2026-07-13 15:01:29] INFO: step=400 {\'loss\': 1.396, \'grad_norm\': 0.811, \'learning_rate\': 5.91e-05, \'epoch\': 1.30}',
|
||||
'[2026-07-13 15:01:34] INFO: Saving checkpoint to /data/checkpoints/medical-cpt-001/checkpoint-400',
|
||||
'[2026-07-13 15:03:37] INFO: step=440 {\'loss\': 1.337, \'grad_norm\': 0.795, \'learning_rate\': 5.35e-05, \'epoch\': 1.43}',
|
||||
'[2026-07-13 15:05:45] INFO: step=480 {\'loss\': 1.286, \'grad_norm\': 0.776, \'learning_rate\': 4.80e-05, \'epoch\': 1.56}',
|
||||
'[2026-07-13 15:07:53] INFO: step=520 {\'loss\': 1.232, \'grad_norm\': 0.758, \'learning_rate\': 4.28e-05, \'epoch\': 1.69}',
|
||||
'[2026-07-13 15:11:59] INFO: step=560 {\'loss\': 1.184, \'grad_norm\': 0.741, \'learning_rate\': 3.79e-05, \'epoch\': 1.82}',
|
||||
'[2026-07-13 15:16:05] INFO: step=590 {\'loss\': 1.146, \'grad_norm\': 0.728, \'learning_rate\': 3.44e-05, \'epoch\': 1.92}',
|
||||
'[2026-07-13 15:16:06] INFO: Training is running normally, estimated remaining time: 00:20:15',
|
||||
]
|
||||
|
||||
const fakeLogLines = [
|
||||
"[2026-02-15 08:30:12] INFO: Loading model from /data/models/qwen2.5-7b",
|
||||
"[2026-02-15 08:30:13] INFO: Loading dataset finance-train-001 (8560 samples)",
|
||||
@@ -591,3 +800,16 @@ export const mockLogContent: LogContent = {
|
||||
size: '2.3 MB',
|
||||
content: fakeLogLines.join('\n'),
|
||||
}
|
||||
|
||||
export const mockTrainingLogContents: Record<string, LogContent> = {
|
||||
'medical-cpt-001_pid28741.log': {
|
||||
file: 'medical-cpt-001_pid28741.log',
|
||||
size: '6.8 MB',
|
||||
content: medicalTrainingLogLines.join('\n'),
|
||||
},
|
||||
'qwen-ft-finance-001_pid12345.log': {
|
||||
file: 'qwen-ft-finance-001_pid12345.log',
|
||||
size: '4.5 MB',
|
||||
content: fakeLogLines.join('\n'),
|
||||
},
|
||||
}
|
||||
|
||||
File diff suppressed because it is too large
Load Diff
@@ -0,0 +1,88 @@
|
||||
<script setup lang="ts">
|
||||
import PageCard from '@/components/PageCard.vue'
|
||||
import { DATASET_TYPE_MAP, STORAGE_MAP, TRAIN_METHOD_MAP, TRAIN_TYPE_MAP } from '@/constants'
|
||||
import type { DatasetItem, FineTuneTask } from '@/types'
|
||||
|
||||
defineProps<{
|
||||
task: FineTuneTask | null
|
||||
dataset: DatasetItem | null
|
||||
baseModelName: string
|
||||
}>()
|
||||
|
||||
function formatDateTime(value?: string) {
|
||||
if (!value) return '-'
|
||||
const date = new Date(value)
|
||||
if (Number.isNaN(date.getTime())) return value
|
||||
return date.toLocaleString('zh-CN', { hour12: false })
|
||||
}
|
||||
</script>
|
||||
|
||||
<template>
|
||||
<PageCard class="task-overview" title="任务信息" subtitle="模型、数据集与运行配置">
|
||||
<div class="overview-layout" aria-label="训练任务信息">
|
||||
<el-descriptions class="task-descriptions task-profile" :column="2" border>
|
||||
<el-descriptions-item label="任务 ID">{{ task?.id ?? '-' }}</el-descriptions-item>
|
||||
<el-descriptions-item label="创建时间">{{ formatDateTime(task?.create_time) }}</el-descriptions-item>
|
||||
<el-descriptions-item label="基座模型">{{ baseModelName }}</el-descriptions-item>
|
||||
<el-descriptions-item label="输出模型">{{ task?.output_model_name || '暂未生成' }}</el-descriptions-item>
|
||||
<el-descriptions-item label="训练方式">
|
||||
{{ task?.train_type ? (TRAIN_TYPE_MAP[task.train_type] || task.train_type) : '未配置' }}
|
||||
</el-descriptions-item>
|
||||
<el-descriptions-item label="训练方法">
|
||||
{{ task?.train_method ? (TRAIN_METHOD_MAP[task.train_method] || task.train_method) : '未配置' }}
|
||||
</el-descriptions-item>
|
||||
<el-descriptions-item label="训练数据集" class-name="dataset-profile">
|
||||
{{ dataset?.name || (task?.train_dataset_id ? '正在加载' : '未配置') }}
|
||||
</el-descriptions-item>
|
||||
<el-descriptions-item label="数据类型">
|
||||
{{ dataset ? (DATASET_TYPE_MAP[dataset.type] || dataset.type || '未配置') : '未配置' }}
|
||||
</el-descriptions-item>
|
||||
<el-descriptions-item label="数据条数">
|
||||
{{ dataset?.count?.toLocaleString('zh-CN') ?? '未配置' }}{{ dataset?.count != null ? ' 条' : '' }}
|
||||
</el-descriptions-item>
|
||||
<el-descriptions-item label="数据大小">{{ dataset?.size || '未配置' }}</el-descriptions-item>
|
||||
<el-descriptions-item label="训练开始时间" class-name="runtime-panel">
|
||||
{{ formatDateTime(task?.create_time) }}
|
||||
</el-descriptions-item>
|
||||
<el-descriptions-item label="训练时长">{{ task?.train_duration || '未配置' }}</el-descriptions-item>
|
||||
<el-descriptions-item label="存储位置">
|
||||
{{ STORAGE_MAP[dataset?.storage_type || ''] || dataset?.storage_type || '未配置' }}
|
||||
</el-descriptions-item>
|
||||
<el-descriptions-item label="使用 GPU">
|
||||
{{ task?.gpus?.length ? task.gpus.join('、') : '未配置' }}
|
||||
</el-descriptions-item>
|
||||
</el-descriptions>
|
||||
</div>
|
||||
</PageCard>
|
||||
</template>
|
||||
|
||||
<style scoped lang="scss">
|
||||
.task-overview {
|
||||
margin-bottom: 0;
|
||||
border: 1px solid #e4e7ed !important;
|
||||
border-radius: 8px !important;
|
||||
box-shadow: none !important;
|
||||
}
|
||||
.overview-layout { width: 100%; }
|
||||
.task-descriptions :deep(.el-descriptions__label) {
|
||||
width: 132px;
|
||||
color: #606266;
|
||||
font-weight: 500;
|
||||
background: #f7f8fa !important;
|
||||
}
|
||||
.task-descriptions :deep(.el-descriptions__content) {
|
||||
color: #303133;
|
||||
font-weight: 500;
|
||||
font-variant-numeric: tabular-nums;
|
||||
}
|
||||
.task-descriptions :deep(.el-descriptions__cell) { padding: 12px 16px !important; }
|
||||
|
||||
@media (max-width: 700px) {
|
||||
.task-descriptions :deep(.el-descriptions__body),
|
||||
.task-descriptions :deep(.el-descriptions__table),
|
||||
.task-descriptions :deep(.el-descriptions__tbody),
|
||||
.task-descriptions :deep(.el-descriptions__row),
|
||||
.task-descriptions :deep(.el-descriptions__cell) { display: block; width: 100%; box-sizing: border-box; }
|
||||
.task-descriptions :deep(.el-descriptions__label) { width: 100%; border-bottom: 0 !important; }
|
||||
}
|
||||
</style>
|
||||
165
frontend/src/views/system/training-log/trainingLogModel.ts
Normal file
165
frontend/src/views/system/training-log/trainingLogModel.ts
Normal file
@@ -0,0 +1,165 @@
|
||||
import type { EChartsOption } from 'echarts'
|
||||
import type { FineTuneTask, TrainingLogFile } from '@/types'
|
||||
|
||||
export interface TrainingMetricData {
|
||||
loss: number[]
|
||||
gradNorm: number[]
|
||||
lr: number[]
|
||||
epoch: number[]
|
||||
}
|
||||
|
||||
export interface TrainingSummary {
|
||||
epoch: string
|
||||
trainLoss: string
|
||||
runtime: string
|
||||
}
|
||||
|
||||
export interface ParsedTrainingLog {
|
||||
metrics: TrainingMetricData
|
||||
summary: TrainingSummary
|
||||
}
|
||||
|
||||
const NUMBER_SOURCE = '[-+]?(?:\\d+(?:\\.\\d*)?|\\.\\d+)(?:[eE][-+]?\\d+)?'
|
||||
|
||||
function escapeRegExp(value: string) {
|
||||
return value.replace(/[.*+?^${}()|[\]\\]/g, '\\$&')
|
||||
}
|
||||
|
||||
function extractNumber(source: string, key: string) {
|
||||
const match = source.match(new RegExp(`['"]?${escapeRegExp(key)}['"]?\\s*:\\s*(${NUMBER_SOURCE})`, 'i'))
|
||||
return match ? Number(match[1]) : undefined
|
||||
}
|
||||
|
||||
function extractSummaryValue(source: string, key: string) {
|
||||
const match = source.match(new RegExp(`['"]?${escapeRegExp(key)}['"]?\\s*(?:=|:)\\s*(${NUMBER_SOURCE})`, 'i'))
|
||||
return match?.[1] || ''
|
||||
}
|
||||
|
||||
/** 根据任务精确选择日志；PID 优先，任务名仅作为明确兜底。 */
|
||||
export function resolveTrainingLogFile(
|
||||
files: TrainingLogFile[],
|
||||
task: Pick<FineTuneTask, 'process_id' | 'name'>,
|
||||
) {
|
||||
const processId = task.process_id
|
||||
if (processId != null) {
|
||||
const pidMatch = files.find((file) => file.pid === processId)
|
||||
if (pidMatch) return pidMatch
|
||||
|
||||
const pidPattern = new RegExp(`(?:^|[^0-9])(?:pid)?${processId}(?:[^0-9]|$)`, 'i')
|
||||
const filenameMatch = files.find((file) => pidPattern.test(file.file))
|
||||
if (filenameMatch) return filenameMatch
|
||||
}
|
||||
|
||||
const taskName = task.name.trim()
|
||||
if (!taskName) return undefined
|
||||
return files.find((file) => file.name.includes(taskName) || file.file.includes(taskName))
|
||||
}
|
||||
|
||||
/** 解析日志中的逐步指标。字段顺序和常见数值格式均不受限制。 */
|
||||
export function parseTrainingMetrics(text: string): TrainingMetricData {
|
||||
const metrics: TrainingMetricData = { loss: [], gradNorm: [], lr: [], epoch: [] }
|
||||
const blocks = text.match(/\{[^{}\r\n]*\}/g) || []
|
||||
|
||||
for (const block of blocks) {
|
||||
const loss = extractNumber(block, 'loss')
|
||||
const gradNorm = extractNumber(block, 'grad_norm')
|
||||
const learningRate = extractNumber(block, 'learning_rate')
|
||||
const epoch = extractNumber(block, 'epoch')
|
||||
if (loss == null || gradNorm == null || learningRate == null) continue
|
||||
metrics.loss.push(loss)
|
||||
metrics.gradNorm.push(gradNorm)
|
||||
metrics.lr.push(learningRate)
|
||||
if (epoch != null) metrics.epoch.push(epoch)
|
||||
}
|
||||
|
||||
return metrics
|
||||
}
|
||||
|
||||
/** 每次都返回新对象，日志截断或切换时不会残留上一轮汇总。 */
|
||||
export function parseTrainingSummary(text: string): TrainingSummary {
|
||||
const emptySummary: TrainingSummary = { epoch: '', trainLoss: '', runtime: '' }
|
||||
const startMatch = /\*{5}\s*train metrics\s*\*{5}/i.exec(text)
|
||||
if (!startMatch) return emptySummary
|
||||
|
||||
const tail = text.slice(startMatch.index + startMatch[0].length)
|
||||
const endMatch = /\*{5}\s*train metrics end\s*\*{5}/i.exec(tail)
|
||||
const body = endMatch ? tail.slice(0, endMatch.index) : tail
|
||||
return {
|
||||
epoch: extractSummaryValue(body, 'epoch'),
|
||||
trainLoss: extractSummaryValue(body, 'train_loss'),
|
||||
runtime: extractSummaryValue(body, 'train_runtime'),
|
||||
}
|
||||
}
|
||||
|
||||
export function parseTrainingLog(text: string): ParsedTrainingLog {
|
||||
return {
|
||||
metrics: parseTrainingMetrics(text),
|
||||
summary: parseTrainingSummary(text),
|
||||
}
|
||||
}
|
||||
|
||||
/** 构建单条训练指标曲线。 */
|
||||
export function buildMetricChartOption(
|
||||
label: string,
|
||||
data: number[],
|
||||
color: string,
|
||||
logScale = false,
|
||||
): EChartsOption {
|
||||
return {
|
||||
grid: { top: 24, right: 20, bottom: 56, left: 56 },
|
||||
tooltip: {
|
||||
trigger: 'axis',
|
||||
axisPointer: { type: 'cross' },
|
||||
backgroundColor: 'rgba(15, 23, 42, 0.9)',
|
||||
borderWidth: 0,
|
||||
textStyle: { color: '#fff', fontSize: 12 },
|
||||
},
|
||||
xAxis: {
|
||||
type: 'category',
|
||||
boundaryGap: false,
|
||||
name: 'Step',
|
||||
nameTextStyle: { color: '#94a3b8', fontSize: 11 },
|
||||
axisLine: { lineStyle: { color: '#e2e8f0' } },
|
||||
axisLabel: { color: '#94a3b8', fontSize: 11 },
|
||||
splitLine: { show: false },
|
||||
},
|
||||
yAxis: {
|
||||
type: logScale ? 'log' : 'value',
|
||||
name: label,
|
||||
nameTextStyle: { color: '#94a3b8', fontSize: 11 },
|
||||
axisLine: { show: false },
|
||||
axisTick: { show: false },
|
||||
axisLabel: { color: '#94a3b8', fontSize: 11 },
|
||||
splitLine: { lineStyle: { color: '#f1f5f9' } },
|
||||
},
|
||||
dataZoom: data.length > 30
|
||||
? [
|
||||
{ type: 'inside', start: 0, end: 100 },
|
||||
{ type: 'slider', height: 16, bottom: 8, borderColor: 'transparent', fillerColor: 'rgba(79,70,229,0.08)', handleStyle: { color: '#4f46e5' } },
|
||||
]
|
||||
: [],
|
||||
series: [
|
||||
{
|
||||
name: label,
|
||||
type: 'line',
|
||||
data,
|
||||
smooth: true,
|
||||
symbol: 'none',
|
||||
lineStyle: { width: 2, color },
|
||||
areaStyle: {
|
||||
color: {
|
||||
type: 'linear',
|
||||
x: 0,
|
||||
y: 0,
|
||||
x2: 0,
|
||||
y2: 1,
|
||||
colorStops: [
|
||||
{ offset: 0, color: `${color}55` },
|
||||
{ offset: 1, color: `${color}05` },
|
||||
],
|
||||
},
|
||||
},
|
||||
},
|
||||
],
|
||||
}
|
||||
}
|
||||
Reference in New Issue
Block a user