docs: fix YAML, broken image links, add library_name for inference

README.md

@@ -85,7 +85,6 @@ The SOCAR Historical Documents AI System is a sophisticated document intelligenc
 - **Ulvi Bashirov** - [GitHub](https://github.com/ulvibashir)
 - **Samir Mehdiyev** - [GitHub](https://github.com/samirmehd1yev)
 
-![Team Performance Results](docs/team_results.png)
 
 | Rank | Team Name | Total Score out of 300 | Total Score out of 500 | Total Presentation Score | Total Score |
 |------|-----------|------------------------|------------------------|--------------------------|-------------|
@@ -215,7 +214,6 @@ We built **3 specialized Jupyter notebooks** to test every component independent
 - `slide_4_summary_table.png` - Complete results table
 - `slide_5_success_rates.png` - CSR vs WSR side-by-side
 
-![OCR Benchmark Results](output/vlm_ocr_benchmark/slide_1_accuracy.png)
 
 **Decision**: **Llama-4-Maverick-17B** selected for OCR endpoint.
 
@@ -477,7 +475,6 @@ We conducted comprehensive benchmarks to select the optimal language model for o
 
 ### Quality Score Comparison
 
-![Quality Score Comparison](output/charts/llm_quality_comparison.png)
 
 **Key Findings**:
 - **GPT-4.1** and **Llama-4-Maverick** tied at **52.0** quality score
@@ -495,7 +492,6 @@ We conducted comprehensive benchmarks to select the optimal language model for o
 
 ### Comprehensive Metrics Breakdown
 
-![Metrics Breakdown](output/charts/llm_metrics_breakdown.png)
 
 **Breakdown by Category**:
 
@@ -519,7 +515,6 @@ We conducted comprehensive benchmarks to select the optimal language model for o
 
 ### Multi-Dimensional Performance Profile
 
-![Radar Profile](output/charts/llm_radar_profile.png)
 
 **Radar Chart Dimensions**:
 
@@ -549,7 +544,6 @@ We conducted comprehensive benchmarks to select the optimal language model for o
 
 ### Response Time Analysis
 
-![Response Time](output/charts/llm_response_time.png)
 
 **Latency Comparison** (Lower is Better):
 
@@ -578,7 +572,6 @@ We conducted comprehensive benchmarks to select the optimal language model for o
 
 ### Complete Overview Dashboard
 
-![Overview Dashboard](output/charts/llm_overview_dashboard.png)
 
 **Four-Panel Analysis**:
 
@@ -609,49 +602,6 @@ We conducted comprehensive benchmarks to select the optimal language model for o
 
 ---
 
-## Live Demo Screenshots
-
-### Web Interface
-
-#### Landing Page
-![Landing Page](output/demo_screenshots/landing%20page.png)
-
-*Main interface with OCR and LLM tabs for document processing and Q&A*
-
-#### LLM Question Answering
-
-![Asking Questions](output/demo_screenshots/asking%20questions%20to%20llm.png)
-
-*Interactive Q&A interface for querying SOCAR historical documents in Azerbaijani*
-
-![LLM Answer Part 2](output/demo_screenshots/llm%20answer%20second%20part.png)
-
-*Complete answer with source citations and document references*
-
-#### OCR Processing
-
-![OCR Started](output/demo_screenshots/excuting%20started%20for%20ocr.png)
-
-*PDF upload and OCR processing initiation*
-
-![OCR Execution](output/demo_screenshots/execution%20ocr%20second%20part.png)
-
-*Real-time OCR processing with page-by-page progress*
-
-![OCR Completed](output/demo_screenshots/execution%20completed%20for%20ocr.png)
-
-*Completed OCR extraction with formatted markdown output*
-
-#### Additional Views
-
-![Second Page](output/demo_screenshots/secondpage.png)
-
-*Document selection and upload interface*
-
-![Third Page](output/demo_screenshots/third%20page.png)
-
-*Advanced features and settings panel*
-
 ---
 
 ## Key Features