OpenConceptLab/ocl_issues#2388 | custom encoder model for reranking

src/components/map-projects/ConfigurationForm.jsx

@@ -32,6 +32,7 @@ import { SCORES_COLOR } from './constants'
 import FilterTable from './FilterTable'
 import MultiAlgoSelector from './MultiAlgoSelector'
 import LookupConfig from './LookupConfig'
+import RerankerConfig from './RerankerConfig'
 
 const VisuallyHiddenInput = styled('input')({
   clip: 'rect(0 0 0 0)',
@@ -46,7 +47,7 @@ const VisuallyHiddenInput = styled('input')({
 });
 
 
-const ConfigurationForm = ({ project, handleFileUpload, file, owner, setOwner, name, setName, description, setDescription, repo, onRepoChange, repoVersion, setRepoVersion, versions, mappedSources, targetSourcesFromRows, algosSelected, setAlgosSelected, sx, algos, validColumns, columns, isValidColumnValue, updateColumn, configure, setConfigure, columnVisibilityModel, setColumnVisibilityModel, onSave, isSaving, candidatesScore, onScoreChange, includeDefaultFilter, setIncludeDefaultFilter, filters, setFilters, locales, isLoadingLocales, setAIAssistantColumns, AIAssistantColumns, inAIAssistantGroup, lookupConfig, setLookupConfig, canBridge, canScispacy }) => {
+const ConfigurationForm = ({ project, handleFileUpload, file, owner, setOwner, name, setName, description, setDescription, repo, onRepoChange, repoVersion, setRepoVersion, versions, mappedSources, targetSourcesFromRows, algosSelected, setAlgosSelected, sx, algos, validColumns, columns, isValidColumnValue, updateColumn, configure, setConfigure, columnVisibilityModel, setColumnVisibilityModel, onSave, isSaving, candidatesScore, onScoreChange, includeDefaultFilter, setIncludeDefaultFilter, filters, setFilters, locales, isLoadingLocales, setAIAssistantColumns, AIAssistantColumns, inAIAssistantGroup, lookupConfig, setLookupConfig, rerankerConfig, setRerankerConfig, isCoreUser, canBridge, canScispacy }) => {
   const { t } = useTranslation();
   const isLLMAlgoNotAllowed = !repoVersion?.match_algorithms?.includes('llm')
   const appliedLocales = filters?.locale ? filters?.locale?.split(',') : []
@@ -222,6 +223,10 @@ const ConfigurationForm = ({ project, handleFileUpload, file, owner, setOwner, n
         onChange={setAlgosSelected}
         repo={repoVersion}
       />
+      {
+        isCoreUser &&
+          <RerankerConfig value={rerankerConfig} onChange={setRerankerConfig} />
+      }
       <>
         <Typography component="div" sx={{fontSize: '16px', fontWeight: 'bold', marginTop: '20px'}}>
           {t('map_project.score_configuration')}

src/components/map-projects/Discuss.jsx

@@ -56,8 +56,11 @@ const Discuss = ({ logs, onAdd }) => {
     }
     if(log.action === 'algo_finished')
       return <>{t('map_project.finished_running')} <b>{log.extras.algo}</b></>
-    if(log.action === 'rerank_finished')
+    if(log.action === 'rerank_finished') {
+      if(log.description)
+        return log.description
       return <>{t('map_project.finished_reranking')}</>
+    }
     return log.description || startCase(log.action)
   }
 

src/components/map-projects/MapProject.jsx

@@ -103,6 +103,7 @@ import ImportToCollection from './ImportToCollection'
 import ProjectLogs from './ProjectLogs';
 import { useAlgos } from './algorithms'
 import AutoMatchDialog from './AutoMatchDialog'
+import { DEFAULT_ENCODER_MODEL } from './rerankerModels'
 
 import './MapProject.scss'
 import '../common/ResizablePanel.scss'
@@ -209,6 +210,7 @@ const MapProject = () => {
   const [analysis, setAnalysis] = React.useState({})
   const [AIModels, setAIModels] = React.useState([])
   const [lookupConfig, setLookupConfig] = React.useState({})
+  const [encoderModel, setEncoderModel] = React.useState(DEFAULT_ENCODER_MODEL)
 
   // import
   const [openImportToCollection, setOpenImportToCollection] = React.useState(false)
@@ -422,6 +424,7 @@ const MapProject = () => {
       setRetired(Boolean(response.data?.include_retired))
       setCandidatesScore(response.data?.score_configuration)
       setLookupConfig(response.data?.lookup_config)
+      setEncoderModel(response.data?.encoder_model || DEFAULT_ENCODER_MODEL)
       setAnalysis(response.data?.analysis || {})
       setProject(response.data)
       setConfigure(false)
@@ -866,6 +869,7 @@ const MapProject = () => {
     formData.append('algorithms', JSON.stringify(map(algosSelected, algo => omit(algo, ['__key']))))
     formData.append('score_configuration', JSON.stringify(candidatesScore))
     formData.append('lookup_config', JSON.stringify(lookupConfig))
+    formData.append('encoder_model', encoderModel)
     formData.append('include_retired', retired)
     formData.append('filters', JSON.stringify(getFilters()))
     const isUpdate = Boolean(project?.id)
@@ -1098,7 +1102,8 @@ const MapProject = () => {
         includeMappings: true,
         mappingBrief: true,
         mapTypes: 'SAME-AS,SAME AS,SAME_AS',
-        reranker: !isMultiAlgo
+        reranker: !isMultiAlgo,
+        ...(encoderModel ? { encoder_model: encoderModel } : {})
       }
 
       forEach(rowBatch, __row => markAlgo(__row.__index, algo.id, 0))
@@ -2040,7 +2045,8 @@ const MapProject = () => {
         limit: algoDef.limit || CANDIDATES_LIMIT,
         offset: offset || 0,
         semantic: ['ocl-semantic', 'custom'].includes(algoDef.type),
-        reranker: !isMultiAlgo && algoDef.provider === 'ocl'
+        reranker: !isMultiAlgo && algoDef.provider === 'ocl',
+        encoder_model: !isMultiAlgo && encoderModel ? encoderModel : undefined
       }).then(response => callback(response, payload))
   }
 
@@ -2201,7 +2207,11 @@ const MapProject = () => {
       markAlgo(index, 'rerank', 0)
       const service = APIService.concepts().appendToUrl('$rerank/')
       try {
-        const response = await service.post({q: query, rows: candidates,});
+        const response = await service.post({
+          q: query,
+          rows: candidates,
+          ...(encoderModel ? { encoder_model: encoderModel } : {})
+        });
 
         setAllCandidates(prev => {
           const newCandidates = {...prev}
@@ -2223,12 +2233,12 @@ const MapProject = () => {
           return newCandidates
         })
         markAlgo(index, 'rerank', 1)
-        log({action: 'rerank_finished'}, index)
+        log({action: 'rerank_finished', description: `Reranked with ${encoderModel}`}, index)
         if(isBulk)
           setTimeout(() => setAutoMatched([index]), 1000)
         return response
       } catch (e) {
-        log({action: 'rerank_failed'}, index)
+        log({action: 'rerank_failed', description: `Rerank failed with ${encoderModel}`}, index)
         markAlgo(index, 'rerank', -2); // optional: failed state
         return null;
       }
@@ -2510,6 +2520,7 @@ const MapProject = () => {
       filters: filters,
       fields_mapped: cols,
       score_configuration: candidatesScore,
+      encoder_model: encoderModel,
       target_repo: repo
     }
   }
@@ -2627,6 +2638,9 @@ const MapProject = () => {
       inAIAssistantGroup={inAIAssistantGroup}
       lookupConfig={lookupConfig}
       setLookupConfig={setLookupConfig}
+      rerankerConfig={encoderModel}
+      setRerankerConfig={setEncoderModel}
+      isCoreUser={isCoreUser}
     />
   )
 

src/components/map-projects/RerankerConfig.jsx

@@ -0,0 +1,97 @@
+import React from 'react'
+import { useTranslation } from 'react-i18next'
+import Autocomplete from '@mui/material/Autocomplete'
+import TextField from '@mui/material/TextField'
+import Collapse from '@mui/material/Collapse'
+import Button from '@mui/material/Button'
+import FormHelperText from '@mui/material/FormHelperText'
+import ListItemText from '@mui/material/ListItemText'
+import UpIcon from '@mui/icons-material/ArrowDropUp';
+import DownIcon from '@mui/icons-material/ArrowDropDown';
+import { CUSTOM_ENCODER_MODEL_OPTION, DEFAULT_ENCODER_MODEL, ENCODER_MODEL_OPTIONS } from './rerankerModels'
+
+const RerankerConfig = ({ value, onChange }) => {
+  const { t } = useTranslation()
+  const [open, setOpen] = React.useState(false)
+  const presetOptions = ENCODER_MODEL_OPTIONS.map(model => ({
+    id: model.id,
+    label: model.description,
+    isDefault: Boolean(model.default),
+    disabled: Boolean(model.disabled)
+  }))
+  const options = [
+    ...presetOptions,
+    { id: CUSTOM_ENCODER_MODEL_OPTION, label: t('map_project.reranker_configuration_custom_option') }
+  ]
+  const isKnownOption = ENCODER_MODEL_OPTIONS.map(option => option.id).includes(value)
+  const selectedOption = isKnownOption ?
+    presetOptions.find(option => option.id === value) :
+    options.find(option => option.id === CUSTOM_ENCODER_MODEL_OPTION)
+  const isCustomSelected = selectedOption?.id === CUSTOM_ENCODER_MODEL_OPTION
+
+  return (
+    <div className='col-xs-12 padding-0' style={{ marginBottom: '16px', marginTop: '4px' }}>
+      <Button
+        size='small'
+        variant='text'
+        color={(value || open) ? 'primary' : 'secondary'}
+        endIcon={open ? <UpIcon fontSize='inherit' /> : <DownIcon fontSize='inherit' />}
+        onClick={() => setOpen(!open)}
+        sx={{ textTransform: 'none' }}
+      >
+        {t('map_project.reranker_configuration')}
+      </Button>
+      <Collapse in={open}>
+        <div className='col-xs-12 padding-0'>
+          <FormHelperText sx={{ marginBottom: '12px', marginTop: '-4px', paddingLeft: '4px' }}>
+            {t('map_project.reranker_configuration_description')}
+          </FormHelperText>
+          <Autocomplete
+            disableClearable
+            options={options}
+            value={selectedOption || null}
+            getOptionLabel={option => option?.id || ''}
+            isOptionEqualToValue={(option, current) => option.id === current.id}
+            getOptionDisabled={(option) => option?.disabled}
+            onChange={(event, option) => {
+              if(option?.id === CUSTOM_ENCODER_MODEL_OPTION) {
+                onChange(isKnownOption ? '' : value)
+                return
+              }
+              onChange(option?.id || DEFAULT_ENCODER_MODEL)
+            }}
+            renderInput={params => (
+              <TextField
+                {...params}
+                label={t('map_project.reranker_configuration_model')}
+                fullWidth
+              />
+            )}
+            renderOption={(props, option) => (
+              <ListItemText
+                {...props}
+                key={option.id}
+              sx={{flexDirection: 'column', alignItems: 'flex-start !important'}}
+              primary={`${option.id}${option.isDefault ? ` (${t('common.default')})` : ''}`}
+              secondary={option.label}
+                />
+            )}
+          />
+          {
+            isCustomSelected &&
+              <TextField
+                sx={{ marginTop: '12px' }}
+                label={t('map_project.reranker_configuration_custom_model')}
+                fullWidth
+                value={value || ''}
+                placeholder={t('map_project.reranker_configuration_placeholder')}
+                onChange={event => onChange(event.target.value || '')}
+              />
+          }
+        </div>
+      </Collapse>
+    </div>
+  )
+}
+
+export default RerankerConfig

src/components/map-projects/rerankerModels.js

@@ -0,0 +1,32 @@
+export const DEFAULT_ENCODER_MODEL = 'BAAI/bge-reranker-v2-m3'
+
+export const ENCODER_MODEL_OPTIONS = [
+  {
+    id: DEFAULT_ENCODER_MODEL,
+    description: 'Multilingual, general-purpose (0.6B)',
+    default: true
+  },
+  {
+    id: 'Qwen/Qwen3-Reranker-0.6B',
+    description: 'General purpose reranker (0.6B)',
+  },
+  {
+    id: 'Qwen/Qwen3-Reranker-4B',
+    description: 'General purpose reranker, best in quality (4B)',
+    disabled: true
+  },
+  {
+    id: 'cross-encoder/ms-marco-MiniLM-L-6-v2',
+    description: 'Fast and lightweight, English-only (23M)',
+  },
+  {
+    id: 'ncbi/MedCPT-Cross-Encoder',
+    description: 'Biomedical domain, trained on PubMed (110M)',
+  },
+  {
+    id: 'Alibaba-NLP/gte-reranker-modernbert-base',
+    description: 'Balanced quality, supports longer descriptions (149M)',
+  },
+]
+
+export const CUSTOM_ENCODER_MODEL_OPTION = '__custom_encoder_model__'

src/i18n/locales/en/translations.json

@@ -559,6 +559,12 @@
     "lookup_configuration_url": "Repository URL",
     "lookup_configuration_token": "Token",
     "lookup_configuration_description": "Configure a Source/CodeSystem for lookup operation to fetch candidates definitions.",
+    "reranker_configuration": "Reranker Configuration",
+    "reranker_configuration_description": "Choose the reranker model used to calculate unified scores for this project. The default model is selected automatically, or you can enter a custom model name.",
+    "reranker_configuration_model": "Reranker model",
+    "reranker_configuration_custom_option": "Custom model name",
+    "reranker_configuration_custom_model": "Custom reranker model",
+    "reranker_configuration_placeholder": "e.g. BAAI/bge-reranker-v2-m3",
     "refresh_candidates_tooltip": "Refresh Candidates",
     "group_candidates": "Group By",
     "sort_candidates": "Sort",

src/i18n/locales/es/translations.json

@@ -534,6 +534,12 @@
     "lookup_configuration_url": "URL del repositorio",
     "lookup_configuration_token": "Token",
     "lookup_configuration_description": "Configure una Fuente/Sistema de códigos para la operación de búsqueda y recuperar definiciones de candidatos.",
+    "reranker_configuration": "Configuracion del reranker",
+    "reranker_configuration_description": "Elija el modelo de reranker utilizado para calcular las puntuaciones unificadas de este proyecto. El modelo predeterminado se selecciona automaticamente, o puede ingresar un nombre de modelo personalizado.",
+    "reranker_configuration_model": "Modelo de reranker",
+    "reranker_configuration_custom_option": "Nombre de modelo personalizado",
+    "reranker_configuration_custom_model": "Modelo de reranker personalizado",
+    "reranker_configuration_placeholder": "p. ej. BAAI/bge-reranker-v2-m3",
     "refresh_candidates_tooltip": "Actualizar candidatos",
     "group_candidates": "Agrupar por",
     "sort_candidates": "Ordenar",

src/i18n/locales/zh/translations.json

@@ -559,6 +559,12 @@
     "lookup_configuration_url": "仓库 URL",
     "lookup_configuration_token": "令牌",
     "lookup_configuration_description": "配置用于查找操作的源/代码系统，以获取候选项定义。",
+    "reranker_configuration": "重排序器配置",
+    "reranker_configuration_description": "为此项目选择用于计算统一分数的重排序器模型。默认模型会自动选中，您也可以输入自定义模型名称。",
+    "reranker_configuration_model": "重排序器模型",
+    "reranker_configuration_custom_option": "自定义模型名称",
+    "reranker_configuration_custom_model": "自定义重排序器模型",
+    "reranker_configuration_placeholder": "例如 BAAI/bge-reranker-v2-m3",
     "refresh_candidates_tooltip": "刷新候选项",
     "group_candidates": "分组依据",
     "sort_candidates": "排序",