d-nerf, 3d vision
NeRF with deformation field.
656 technical terms and definitions
NeRF with deformation field.
Minimize variance of parameter estimates.
D-vector is a deep neural network embedding that represents speaker characteristics for verification and multi-speaker speech synthesis.
Variation between dies on same wafer.
Discriminator-Actor-Critic combines adversarial imitation learning with off-policy RL improving sample efficiency.
Iterative imitation learning.
Dagster is data orchestration with software-defined assets. Type-safe pipelines.
DALL-E 3 improves text-to-image through enhanced caption generation and refined training.
DALL-E tokenizer uses discrete visual codes for autoregressive image generation from text.
# Daly City, California ## Overview Daly City is a densely populated city located in **San Mateo County, California**, immediately south of San Francisco. It serves as a critical residential and commercial hub in the San Francisco Bay Area. ## Basic Information - **Official Name:** City of Daly City - **County:** San Mateo County - **State:** California, USA - **Incorporated:** March 22, 1911 - **Named After:** John Donald Daly (1841–1923), a dairy farmer and landowner ## Geographic Data ### Location Coordinates $$ \text{Latitude: } 37.6879° \, N $$ $$ \text{Longitude: } 122.4702° \, W $$ ### Area Calculations - **Total Area:** $7.7 \, \text{mi}^2$ (approximately $19.9 \, \text{km}^2$) - **Land Area:** $7.6 \, \text{mi}^2$ - **Water Area:** $0.1 \, \text{mi}^2$ Area conversion formula: $$ A_{\text{km}^2} = A_{\text{mi}^2} \times 2.58999 $$ Example: $$ 7.7 \, \text{mi}^2 \times 2.58999 = 19.94 \, \text{km}^2 $$ ### Elevation - **Average Elevation:** $\approx 520 \, \text{ft}$ ($158 \, \text{m}$) - **Highest Point:** $\approx 925 \, \text{ft}$ ($282 \, \text{m}$) Elevation conversion: $$ h_{\text{meters}} = h_{\text{feet}} \times 0.3048 $$ ## Population Statistics ### Census Data | Year | Population | Change | |------|------------|--------| | 2000 | 103,621 | — | | 2010 | 101,123 | $-2.4\%$ | | 2020 | 104,901 | $+3.7\%$ | ### Population Density $$ \rho = \frac{P}{A} $$ Where: - $\rho$ = population density - $P$ = population - $A$ = area Calculation: $$ \rho = \frac{104,901}{7.7 \, \text{mi}^2} \approx 13,623 \, \text{people/mi}^2 $$ In metric: $$ \rho_{\text{metric}} = \frac{104,901}{19.94 \, \text{km}^2} \approx 5,261 \, \text{people/km}^2 $$ ## Demographics ### Ethnic Composition (Approximate) - **Asian:** $\approx 56\%$ - Filipino: $\approx 33\%$ (largest subgroup) - Chinese: $\approx 15\%$ - Other Asian: $\approx 8\%$ - **White:** $\approx 18\%$ - **Hispanic/Latino:** $\approx 20\%$ - **Black/African American:** $\approx 3\%$ - **Other/Mixed:** $\approx 3\%$ ### Diversity Index The Simpson's Diversity Index can be calculated as: $$ D = 1 - \sum_{i=1}^{n} p_i^2 $$ Where $p_i$ is the proportion of each ethnic group. For Daly City: $$ D = 1 - (0.56^2 + 0.18^2 + 0.20^2 + 0.03^2 + 0.03^2) $$ $$ D = 1 - (0.3136 + 0.0324 + 0.04 + 0.0009 + 0.0009) $$ $$ D = 1 - 0.3878 = 0.6122 $$ A diversity index of $D \approx 0.61$ indicates **moderate-to-high diversity**. ## Climate Data ### Climate Classification - **Köppen Classification:** $Csb$ (Mediterranean Climate / Warm-summer Mediterranean) ### Temperature Statistics | Metric | Value (°F) | Value (°C) | |--------|------------|------------| | Average High | $63°F$ | $17.2°C$ | | Average Low | $49°F$ | $9.4°C$ | | Annual Mean | $56°F$ | $13.3°C$ | Temperature conversion formula: $$ T_C = \frac{5}{9}(T_F - 32) $$ ### Precipitation - **Annual Rainfall:** $\approx 20 \, \text{inches}$ ($508 \, \text{mm}$) Conversion: $$ P_{\text{mm}} = P_{\text{in}} \times 25.4 $$ ### Fog Frequency Daly City is known for its frequent fog due to: $$ \text{Fog Formation} \propto \frac{\Delta T \times H}{W} $$ Where: - $\Delta T$ = temperature differential (ocean vs. land) - $H$ = humidity - $W$ = wind speed The city experiences fog approximately **100+ days per year**. ## Transportation ### BART Stations - **Daly City Station** (opened 1973) - **Colma Station** (opened 1996) ### Commute Statistics Average commute time: $$ \bar{t} \approx 32 \, \text{minutes} $$ Modal split (approximate): - **Drive Alone:** $\approx 55\%$ - **Public Transit:** $\approx 25\%$ - **Carpool:** $\approx 12\%$ - **Walk/Bike/Other:** $\approx 8\%$ ## Economic Data ### Median Household Income $$ \tilde{Y} \approx 95{,}000\,\text{ USD (2023 estimate)} $$ ### Housing Statistics - **Median Home Price:** approx USD 950,000 - **Median Rent:** approx USD 2,500/month Housing affordability ratio: $$ \text{Affordability Ratio} = \frac{\text{Median Home Price}}{\text{Median Income}} = \frac{950,000}{95,000} \approx 10 $$ A ratio $> 5$ indicates a **severely unaffordable** housing market. ## Cultural Significance ### "Little Manila" Daly City has one of the **highest concentrations of Filipino Americans** in the United States. Filipino population percentage: $$ P_{\text{Filipino}} = \frac{\text{Filipino Residents}}{\text{Total Population}} \times 100 \approx 33\% $$ ### Historical Reference: "Little Boxes" The song *"Little Boxes"* (1962) by **Malvina Reynolds** was inspired by the **Westlake neighborhood** tract housing. Verse structure analysis (syllables per line): $$ \text{Meter Pattern} = \{8, 8, 8, 6\} \quad \text{(approximate)} $$ ## Key Landmarks - **Westlake Shopping Center** — One of the first planned shopping centers in the U.S. - **Serramonte Center** — Major regional shopping mall - **Top of the Hill Park** — Scenic overlook - **Thornton State Beach** — Coastal access point ## Government ### City Council Structure - **Council Members:** 5 - **Mayor:** Elected by council (rotates annually) ### Budget Formula (Simplified) $$ B_{\text{total}} = R_{\text{tax}} + R_{\text{fees}} + G_{\text{state}} + G_{\text{federal}} $$ Where: - $B_{\text{total}}$ = total budget - $R_{\text{tax}}$ = tax revenue - $R_{\text{fees}}$ = fees and permits - $G_{\text{state}}$ = state grants - $G_{\text{federal}}$ = federal grants ## Distance Calculations ### Distance from Key Locations | Destination | Distance (mi) | Distance (km) | |-------------|---------------|---------------| | San Francisco (Downtown) | $8$ | $12.9$ | | San Jose | $40$ | $64.4$ | | Oakland | $20$ | $32.2$ | | SFO Airport | $5$ | $8.0$ | Haversine formula for great-circle distance: $$ d = 2r \arcsin\left(\sqrt{\sin^2\left(\frac{\Delta\phi}{2}\right) + \cos(\phi_1)\cos(\phi_2)\sin^2\left(\frac{\Delta\lambda}{2}\right)}\right) $$ Where: - $r$ = Earth's radius ($\approx 3,959 \, \text{mi}$ or $6,371 \, \text{km}$) - $\phi_1, \phi_2$ = latitudes - $\Delta\phi$ = difference in latitude - $\Delta\lambda$ = difference in longitude ## Reference | Category | Value | |----------|-------| | **Population** | $\approx 104,901$ | | **Area** | $7.7 \, \text{mi}^2$ | | **Density** | $\approx 13,623 \, \text{/mi}^2$ | | **Elevation** | $\approx 520 \, \text{ft}$ | | **Founded** | 1911 | | **Filipino %** | $\approx 33\%$ | | **Median Income** | approx USD 95,000 | | **Climate** | $Csb$ (Mediterranean) |
Pattern trenches fill with metal CMP to remove excess.
Type of jailbreak prompt claiming to remove restrictions.
Jailbreak technique.
Adversarial training for domain adaptation.
Sparsify then merge models.
Information in soft targets beyond hard labels.
Dark knowledge is information in teacher's soft predictions beyond hard labels aiding distillation.
Scatter light off defects for enhanced detection.
Differentiable Architecture Search enables gradient-based NAS by relaxing the discrete architecture search space into continuous representations.
Dask parallelizes pandas/numpy across cores or cluster. Lazy evaluation. Scale out analysis.
Process of labeling data for training or evaluation.
Data anonymization removes identifying information preventing re-identification.
Remove or obscure identifying information.
Privacy-preserving augmentation creates variations without exposing original data.
Transform existing data (paraphrase translate corrupt) to increase training diversity.
Data cards describe dataset characteristics collection and potential biases.
Documentation of dataset characteristics and collection.
Groups of data appearing together.
Automatically gather process data and metrology results.
Check if test data leaked into training set.
When test data appears in training data inflating scores.
Remove duplicate training examples.
Remove duplicate examples from dataset.
When input data distribution changes over time.
How much data ViT needs.
LLMs extract structured data from unstructured text: invoices, resumes, contracts. Output JSON for downstream use.
Remove low-quality data.
Remove low-quality or irrelevant examples.
I can explain data labeling strategies, guidelines, QC loops, and how to design good annotation instructions.
When model exposes training data or sensitive information.
Efficient data feeding to GPUs.
Data minimization collects only necessary information reducing privacy risks.
Data mix balances domains (web, books, code). Proportions affect model capabilities. Code improves reasoning.
Combine different data sources.
Impact of data sequence.
Data parallel: same model, different data. Model parallel: split model across GPUs. Hybrid for largest models.
Replicate model on each device process different data batches.
Data pipelines (Airflow, Dagster) orchestrate ETL for training data. Reliable, versioned, scheduled.
Data poisoning corrupts training data to degrade model performance or insert vulnerabilities.
Inject malicious data into training set to corrupt model behavior.