deepearth/central_florida_native_plants

Central Florida Native Plants Language Embeddings

This dataset contains language embeddings for 232 native plant species from Central Florida, extracted using the DeepSeek-V3 language model.

Dataset Summary

This dataset provides pre-computed language embeddings for Central Florida plant species. Each species has been encoded using the prompt "Ecophysiology of {species_name}:" to capture semantic information about the plant's ecological characteristics.

Dataset Structure

Data Instances

Each species is represented by:

• A PyTorch file (.pt) containing a dictionary with embeddings and metadata
• A CSV file containing the token mappings

Embedding File Structure

Each .pt file contains a dictionary with:

• mean_embedding: Tensor of shape [7168] - mean-pooled embedding across all tokens (including prompt)
• token_embeddings: Tensor of shape [num_tokens, 7168] - individual token embeddings
• species_name: String - the species name
• taxon_id: String - GBIF taxon ID
• num_tokens: Integer - number of tokens (typically 18-20)
• embedding_stats: Dictionary with embedding statistics
• timestamp: String - when the embedding was created

Dataset Viewer Structure

The Parquet files in the dataset viewer contain:

• taxon_id: GBIF taxonomic identifier
• species_name: Scientific name of the plant species
• timestamp: When the embedding was created
• token_position: Position of token in sequence
• token_id: Token ID in model vocabulary
• token_str: String representation of token
• is_species_token: Whether this token is part of the species name
• token_embedding: 7168-dimensional embedding vector for this specific token
• species_mean_embedding: 7168-dimensional mean embedding of species name tokens only
• all_tokens_mean_embedding: 7168-dimensional mean embedding across all tokens (including prompt)
• num_tokens: Total number of tokens for this species
• num_species_tokens: Number of tokens that are part of the species name

Token Mapping Structure

Token mapping CSV files contain:

• position: Token position in sequence
• token_id: Token ID in model vocabulary
• token: Token string representation

Data Splits

This dataset contains a single split with embeddings for all 232 species.

Important Note on Embeddings

This dataset provides two types of mean embeddings:

1. species_mean_embedding (in dataset viewer): The mean embedding calculated from ONLY the tokens that represent the species name itself. This provides a more focused representation of the species.

2. all_tokens_mean_embedding or mean_embedding (in .pt files): The mean embedding calculated from ALL tokens in the prompt, including "Ecophysiology of", the species name, and the ":" token. This is the original embedding as extracted from the model.

For most use cases, species_mean_embedding is recommended as it captures the semantic representation of the species name without the influence of the prompt template.

Dataset Creation

Model Information

• Model: DeepSeek-V3-0324-UD-Q4_K_XL
• Parameters: 671B (4.5-bit quantized GGUF format)
• Embedding Dimension: 7168
• Context: 2048 tokens
• Prompt Template: "Ecophysiology of {species_name}:"

Source Data

Species names are based on GBIF (Global Biodiversity Information Facility) taxonomy for plants native to Central Florida.

Usage

Loading Embeddings

import torch
import pandas as pd
from huggingface_hub import hf_hub_download

# Download a specific embedding
repo_id = "deepearth/central_florida_native_plants"
species_id = "2650927"  # Example GBIF ID

# Download embedding file
embedding_path = hf_hub_download(
    repo_id=repo_id,
    filename=f"embeddings/{species_id}.pt",
    repo_type="dataset"
)

# Load embedding dictionary
data = torch.load(embedding_path)

# Access embeddings
mean_embedding = data['mean_embedding']  # Shape: [7168] - mean of all tokens
token_embeddings = data['token_embeddings']  # Shape: [num_tokens, 7168]
species_name = data['species_name']

print(f"Species: {species_name}")
print(f"Mean embedding shape: {mean_embedding.shape}")
print(f"Token embeddings shape: {token_embeddings.shape}")

# For species-only mean embedding, use the dataset viewer or compute from species tokens
# The dataset viewer provides 'species_mean_embedding' which is the mean of only
# the tokens that are part of the species name (excluding prompt tokens)

# Download and load token mapping
token_path = hf_hub_download(
    repo_id=repo_id,
    filename=f"tokens/{species_id}.csv",
    repo_type="dataset"
)
tokens = pd.read_csv(token_path)

Batch Download

from huggingface_hub import snapshot_download

# Download entire dataset
local_dir = snapshot_download(
    repo_id="deepearth/central_florida_native_plants",
    repo_type="dataset",
    local_dir="./florida_plants"
)

Additional Information

Dataset Curators

This dataset was created by the DeepEarth Project to enable machine learning research on biodiversity and ecology.

Licensing Information

This dataset is licensed under the MIT License.

Citation Information

@dataset{deepearth_florida_plants_2025,
  title={Central Florida Native Plants Language Embeddings},
  author={DeepEarth Project},
  year={2025},
  publisher={Hugging Face},
  howpublished={\url{https://huggingface.co/datasets/deepearth/central_florida_native_plants}}
}

Contributions

Thanks to @legel for creating this dataset.