## Line Chart: Exact Match vs. SFT Data Ratio ### Overview The image is a line chart comparing the "Exact Match (%)" against the "SFT Data Ratio (×10-4)" for four different categories: ID, CMP, POOD, and OOD. The chart illustrates how the exact match percentage changes as the SFT data ratio increases for each category. ### Components/Axes * **Y-axis (Vertical):** "Exact Match (%)", ranging from 0 to 100 in increments of 20. * **X-axis (Horizontal):** "SFT Data Ratio (×10-4)", ranging from 0 to 6 in increments of 1. * **Legend (Right):** * *ID*: Light blue dotted line with diamond markers. * *CMP*: Light purple solid line with square markers. * *POOD*: Light red dashed line with circle markers. * *OOD*: Light teal dash-dotted line with triangle markers. ### Detailed Analysis * **ID (Light Blue, Dotted, Diamond Markers):** The "ID" line remains consistently at 100% across all SFT Data Ratio values. * (0, 100) * (1, 100) * (2, 100) * (3, 100) * (4, 100) * (5, 100) * (6, 100) * **CMP (Light Purple, Solid, Square Markers):** The "CMP" line increases sharply from approximately 0% to approximately 80% between SFT Data Ratio values of 0 and 1. It then continues to increase, reaching approximately 95% at a ratio of 2, and plateaus near 100% after a ratio of 4. * (0, 0) * (1, 80) * (2, 95) * (3, 98) * (4, 99) * (5, 100) * (6, 100) * **POOD (Light Red, Dashed, Circle Markers):** The "POOD" line increases from approximately 0% to approximately 50% between SFT Data Ratio values of 0 and 1. It continues to increase, reaching approximately 90% at a ratio of 3, and plateaus near 100% after a ratio of 5. * (0, 0) * (1, 50) * (2, 75) * (3, 90) * (4, 95) * (5, 98) * (6, 100) * **OOD (Light Teal, Dash-Dotted, Triangle Markers):** The "OOD" line increases gradually from approximately 0% to approximately 50% between SFT Data Ratio values of 0 and 2. It continues to increase, reaching approximately 90% at a ratio of 5, and plateaus near 90% after a ratio of 5. * (0, 0) * (1, 10) * (2, 45) * (3, 70) * (4, 85) * (5, 90) * (6, 90) ### Key Observations * The "ID" category consistently achieves a 100% exact match, regardless of the SFT Data Ratio. * The "CMP" category shows the most rapid improvement in exact match percentage with increasing SFT Data Ratio. * The "OOD" category shows the slowest improvement in exact match percentage with increasing SFT Data Ratio. * All categories except "ID" show a positive correlation between SFT Data Ratio and Exact Match (%). ### Interpretation The chart suggests that increasing the SFT Data Ratio generally improves the exact match percentage for the "CMP", "POOD", and "OOD" categories. The "ID" category, however, maintains a perfect match regardless of the SFT Data Ratio, implying it may be inherently easier to match or is already optimized. The "CMP" category benefits the most from increased SFT data, while "OOD" benefits the least. This could indicate differences in the complexity or characteristics of the data within each category. The data suggests that "CMP" and "POOD" benefit from increased SFT data ratio, but eventually plateau, suggesting diminishing returns. "OOD" continues to improve with increased SFT data ratio, but at a slower rate.

## Chart: Exact Match vs. SFT Data Ratio ### Overview The image presents a line chart illustrating the relationship between the SFT (Supervised Fine-Tuning) Data Ratio and the Exact Match percentage for four different models/metrics: ID, CMP, POOD, and OOD. The chart demonstrates how the performance (Exact Match) of each model improves as the amount of SFT data increases. ### Components/Axes * **X-axis:** SFT Data Ratio (×10⁻⁴). Scale ranges from approximately 0 to 6. * **Y-axis:** Exact Match (%). Scale ranges from 0 to 100. * **Legend:** Located in the top-right corner. Contains the following labels and corresponding colors: * ID (Blue, dotted line) * CMP (Purple, solid line) * POOD (Orange, dashed line) * OOD (Teal, dash-dot line) * **Gridlines:** Present to aid in reading values. ### Detailed Analysis * **ID (Blue, dotted):** The line starts at approximately 100% Exact Match at an SFT Data Ratio of 0. It remains relatively stable at around 100% throughout the entire range of SFT Data Ratios. * (0, 100) * (1, 100) * (2, 100) * (3, 100) * (4, 100) * (5, 100) * (6, 100) * **CMP (Purple, solid):** The line begins at approximately 40% Exact Match at an SFT Data Ratio of 0. It rapidly increases to approximately 95% at an SFT Data Ratio of 1. It plateaus around 95-98% for the remainder of the SFT Data Ratio range. * (0, 40) * (1, 95) * (2, 98) * (3, 98) * (4, 98) * (5, 98) * (6, 98) * **POOD (Orange, dashed):** The line starts at approximately 0% Exact Match at an SFT Data Ratio of 0. It increases sharply to approximately 85% at an SFT Data Ratio of 2. It continues to increase, but at a slower rate, reaching approximately 95% at an SFT Data Ratio of 6. * (0, 0) * (1, 40) * (2, 85) * (3, 90) * (4, 93) * (5, 95) * (6, 95) * **OOD (Teal, dash-dot):** The line begins at approximately 0% Exact Match at an SFT Data Ratio of 0. It steadily increases, reaching approximately 75% at an SFT Data Ratio of 6. This line exhibits the slowest rate of increase among the four models. * (0, 0) * (1, 15) * (2, 40) * (3, 60) * (4, 75) * (5, 80) * (6, 85) ### Key Observations * The ID model consistently achieves 100% Exact Match regardless of the SFT Data Ratio. * CMP demonstrates the fastest improvement in Exact Match with increasing SFT Data Ratio, reaching a plateau quickly. * POOD and OOD show a more gradual improvement, with POOD outperforming OOD across all SFT Data Ratios. * OOD requires the most SFT data to achieve a reasonable level of Exact Match. ### Interpretation The chart suggests that increasing the amount of SFT data generally improves the performance of the models, as measured by Exact Match. However, the extent of improvement varies significantly between models. The ID model appears to be already highly optimized or inherently capable of achieving perfect matches, while CMP benefits the most from SFT. POOD and OOD show a more moderate improvement, indicating they may require more data or different optimization strategies to reach the same level of performance as ID and CMP. The differences in performance between the models could be due to variations in their architectures, training data, or the specific tasks they are designed to perform. The chart highlights the importance of SFT in enhancing model accuracy and the potential for diminishing returns as the SFT Data Ratio increases. The fact that ID remains at 100% suggests it may be a baseline or a model that has already reached its maximum potential.

## Line Chart: Exact Match Performance vs. SFT Data Ratio ### Overview This image is a line chart comparing the performance of four different methods or models (ID, CMP, POOD, OOD) on an "Exact Match" metric as a function of the "SFT Data Ratio." The chart demonstrates how each method's accuracy improves with increasing amounts of supervised fine-tuning (SFT) data. ### Components/Axes * **Chart Type:** Multi-series line chart. * **Y-Axis:** * **Label:** "Exact Match (%)" * **Scale:** Linear, from 0 to 100, with major tick marks every 20 units (0, 20, 40, 60, 80, 100). * **X-Axis:** * **Label:** "SFT Data Ratio (×10⁻⁴)" * **Scale:** Linear, from 0 to 6, with major tick marks at every integer (0, 1, 2, 3, 4, 5, 6). The "(×10⁻⁴)" indicates the values are scaled; for example, "1" represents a ratio of 0.0001. * **Legend:** Located in the bottom-right quadrant of the chart area. It contains four entries: 1. **ID:** Blue dotted line with diamond markers (♦). 2. **CMP:** Purple solid line with square markers (■). 3. **POOD:** Orange dashed line with circle markers (●). 4. **OOD:** Green dash-dot line with triangle markers (▲). ### Detailed Analysis The chart plots the "Exact Match (%)" for each method at discrete "SFT Data Ratio" points. The following data points are approximate visual estimates: **1. ID (Blue, Dotted, Diamonds):** * **Trend:** Perfectly flat, horizontal line at the top of the chart. * **Data Points:** Maintains 100% Exact Match across all SFT Data Ratios from 0 to 6. **2. CMP (Purple, Solid, Squares):** * **Trend:** Steep, rapid ascent from near 0% to near 100%, followed by a plateau. * **Data Points:** * Ratio 0: ~0% * Ratio 0.5: ~40% * Ratio 1: ~80% * Ratio 1.5: ~95% * Ratio 2: ~98% * Ratios 3-6: Plateaus at ~100%. **3. POOD (Orange, Dashed, Circles):** * **Trend:** Steady, strong upward curve that approaches 100% more gradually than CMP. * **Data Points:** * Ratio 0: ~0% * Ratio 0.5: ~22% * Ratio 1: ~50% * Ratio 1.5: ~72% * Ratio 2: ~88% * Ratio 3: ~95% * Ratio 4: ~97% * Ratios 5-6: Plateaus near ~99%. **4. OOD (Green, Dash-Dot, Triangles):** * **Trend:** Slower, more gradual ascent that begins to plateau at a lower level than the other methods. * **Data Points:** * Ratio 0: ~0% * Ratio 1: ~15% * Ratio 2: ~45% * Ratio 3: ~70% * Ratio 4: ~85% * Ratio 5: ~90% * Ratio 6: ~90% (Plateaus). ### Key Observations 1. **Performance Hierarchy:** At all non-zero data ratios, the performance order is consistent: ID (best) > CMP > POOD > OOD (worst). 2. **Data Efficiency:** CMP is the most data-efficient method among the three that start at 0%. It reaches near-perfect performance with a very small data ratio (≈1.5 x 10⁻⁴). 3. **Ceiling Effect:** The ID method appears to be a theoretical or ideal baseline, as it shows perfect performance regardless of data ratio. 4. **OOD Limitation:** The OOD method not only learns slower but also appears to hit a performance ceiling around 90%, suggesting a fundamental limitation in its ability to achieve exact matches compared to the other approaches. 5. **Convergence:** Both CMP and POOD converge to near 100% performance, but CMP requires significantly less data to get there. ### Interpretation This chart likely illustrates a study on model robustness or generalization, comparing in-distribution (ID) performance against various out-of-distribution (OOD) or specialized training scenarios (CMP, POOD). * **What the data suggests:** The "ID" line represents the upper-bound performance on familiar data. The other lines show how different training or evaluation strategies (CMP, POOD, OOD) recover this performance as they are exposed to more fine-tuning data. The steep rise of CMP suggests it is a highly effective strategy for quickly adapting to or handling a specific distribution shift. The slower rise and lower plateau of OOD indicate it represents a more challenging distribution shift that the model cannot fully overcome with additional data alone. * **Relationship between elements:** The X-axis (data ratio) is the independent variable controlling the amount of adaptation. The Y-axis (exact match) is the dependent measure of success. The diverging paths of the lines highlight the varying difficulty of the tasks or distributions they represent. * **Notable anomaly:** The perfect, flat line for "ID" is striking. It implies that for the in-distribution test set, the base model (with zero additional SFT data) already achieves 100% exact match, or that this line represents a different, non-adaptive benchmark. This serves as a control, emphasizing that the challenges shown for the other lines are due to distribution shift, not model incapability on the core task.

## Line Graph: Exact Match Performance vs. SFT Data Ratio ### Overview The graph illustrates the relationship between the SFT Data Ratio (scaled by ×10⁻⁴) and the Exact Match (%) performance of four distinct methods: ID, CMP, POOD, and OOD. Performance is measured as a percentage, with all methods plateauing near 100% at higher data ratios. ### Components/Axes - **X-axis**: SFT Data Ratio (×10⁻⁴), ranging from 0 to 6 in increments of 1. - **Y-axis**: Exact Match (%), ranging from 0 to 100% in increments of 20. - **Legend**: Located in the bottom-right corner, mapping: - Blue diamonds (dashed line): ID - Purple squares (solid line): CMP - Red circles (dashed line): POOD - Green triangles (dashed line): OOD ### Detailed Analysis 1. **ID (Blue Diamonds)**: - Starts at 0% at x=0. - Rises sharply to 100% by x=1. - Remains flat at 100% for x ≥ 1. 2. **CMP (Purple Squares)**: - Begins at 0% at x=0. - Increases gradually, reaching 100% by x=2. - Plateaus at 100% for x ≥ 2. 3. **POOD (Red Circles)**: - Starts at 0% at x=0. - Rises steadily, surpassing CMP around x=2. - Reaches 100% by x=4 and plateaus. 4. **OOD (Green Triangles)**: - Begins at 0% at x=0. - Increases slowly, reaching ~85% by x=6. - Stabilizes near 85% for x ≥ 6. ### Key Observations - **Early Performance**: ID achieves 100% performance fastest (x=1), followed by CMP (x=2) and POOD (x=4). - **Late-Stage Growth**: OOD lags significantly, only reaching ~85% by x=6. - **Crossing Trends**: POOD overtakes CMP between x=2 and x=3, indicating superior performance at higher data ratios. - **Plateaus**: All methods plateau near 100% except OOD, which plateaus at ~85%. ### Interpretation The graph suggests that **ID** is the most efficient method, achieving full performance with minimal data. **CMP** and **POOD** show similar trajectories but diverge at higher ratios, with POOD outperforming CMP beyond x=3. **OOD** demonstrates the weakest performance, requiring the largest data ratio to reach ~85% and failing to match the others. This could reflect differences in algorithmic efficiency, data utilization, or model architecture. The plateauing behavior implies diminishing returns beyond certain data ratios, highlighting a potential threshold for optimal performance.