Page Overview
- Client industry: SMT mounting and electronic assembly
- Business direction: Quality problem traceability and defect analysis
- Capability area: Model access and configuration
What the data for this category looks like
Business data sources for SMT mounting and electronic assembly include customer-provided BOM, Gerber files, coordinate files, MES systems, mounters, SPI/AOI inspection equipment, reflow oven temperature records, first article confirmation reports, maintenance records, and traceability labels. Data updates occur in real time per work order and each line change. Documents include structured part numbers, reference designators, component batch information, plus unstructured furnace temperature curves and defect image reports. Fields include millimeter-level mounting coordinates, Celsius-grade furnace temperature parameters and other professional units.
Constraints on model access and configuration
The work order granularity of this category’s data is fine, updates are real-time, and the data includes both structured fields and unstructured images. Model access configuration must adapt to requirements for short-context fast response, multimodal data processing, and accurate field matching. API interfaces for multiple data sources need independent access credential configuration to avoid permission confusion. The data length of a single work order fluctuates greatly, so context and output length must be limited to avoid call timeouts. Processing of visual defect reports must enable corresponding support to ensure extraction of defect features such as solder joint offset and dry joint.
How to set the configuration
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
maxContext | 800–1200 token | Adapts to the total length of BOM, line change records, and defect reports for a single work order, avoiding context overflow |
maxTokens | 600–800 token | Limits the output length of a single round of analysis, matching the concise requirements of SMT defect analysis |
vision | Enabled | Adapts to SPI/AOI image defect reports, extracting visual features such as solder joint offset and dry joint |
temperature | 0.1–0.3 | Reduces randomness, ensuring consistent matching of key fields such as part numbers and reference designators |
AIPROXY_API_ENDPOINT | Configure corresponding endpoints per data source | Connects to API interfaces of MES, mounters, and AOI equipment to pull data from multiple sources |
AIPROXY_API_TOKEN | Generate exclusive tokens per data source | Isolates access permissions for different devices and systems, avoiding configuration conflicts |
Three Common Configuration Mistakes
- Symptom: When adding an embedding model with the same name, existing configuration is automatically overwritten. Cause: The platform uses the model name as the unique identifier by default. Overwriting triggers when a custom model alias is not configured.
- Symptom: After the knowledge base returns results, output content includes the phrase "Reference Marker: [1]". Cause: Knowledge base reference marker generation rules have not been adjusted. Redundant marker text is appended by default.
- Symptom: A context length limit exceeded error occurs when initiating a defect analysis request. Cause: The
maxContextparameter has not been adjusted based on single work order data volume. Total incoming data length exceeds the configured limit.
How to Confirm Configuration Is Complete
- Upload the BOM and SPI/AOI reports for a single work order, check whether the model correctly extracts part numbers, reference designators, and defect types.
- Attempt to add multiple models with the same name, verify whether an overwrite prompt triggers or existing configurations are retained (custom model aliases must be configured in advance).
- Initiate a complete quality traceability analysis request, check whether output content contains no redundant reference markers.
- View FastGPT v4.17.0 API call logs, confirm that requests using
AIPROXY_API_ENDPOINTandAIPROXY_API_TOKENsuccessfully pull data from corresponding data sources.
The values provided are common starting points and should be measured against the reader's own samples.