magnet.utils package

Subpackages

magnet.utils.mlx package
- Submodules
- magnet.utils.mlx.mistral module

Submodules

magnet.utils.data_classes module

class magnet.utils.data_classes.AskParameters(m: str = 'mistralai/Mistral-7B-Instruct-v0.1', q: str = 'What is your itinerary?', t: float = 1.0, n: int = 8096, p: str = 'qa_ref', cb: Optional[Callable] = None, docs: List[str] = <factory>, context: str = '', vllm: bool = False)[source]

Bases: object

cb: Callable | None = None

context: str = ''

docs: List[str]

m: str = 'mistralai/Mistral-7B-Instruct-v0.1'

n: int = 8096

p: str = 'qa_ref'

q: str = 'What is your itinerary?'

t: float = 1.0

vllm: bool = False

class magnet.utils.data_classes.EmbeddingPayload(document: str, embedding: list, text: list, model: str)[source]

Bases: object

Represents a payload for embedding text data.

Attributes:: document (str): The document associated with the text data. embedding (list): The embedding of the text data. text (list): The text of the data. model (str): The model used for embedding the text data.

document: str

embedding: list

model: str

text: list

class magnet.utils.data_classes.GeneratedPayload(query: str, prompt: str, context: list, result: str, model: str)[source]

Bases: object

Represents a payload generated by a system.

Args:: query (str): The query associated with the payload. prompt (str): The prompt associated with the payload. context (list): The context associated with the payload. result (str): The result generated by the system. model (str): The model used to generate the payload.

context: list

model: str

prompt: str

query: str

result: str

class magnet.utils.data_classes.IndexConfig(milvus_uri: Optional[str] = None, milvus_port: Optional[int] = None, milvus_user: Optional[str] = None, milvus_password: Optional[str] = None, dimension: Optional[int] = None, model: Optional[str] = None, name: Optional[str] = None, options: Dict[Optional[dict], Any] = <factory>)[source]

Bases: object

dimension: int | None = None

milvus_password: str | None = None

milvus_port: int | None = None

milvus_uri: str | None = None

milvus_user: str | None = None

model: str | None = None

name: str | None = None

options: Dict[dict | None, Any]

class magnet.utils.data_classes.JobParams(milvus_host: str, milvus_port: int, milvus_username: str, milvus_password: str, milvus_collection: str, nats_host: str, nats_username: str, nats_password: str, nats_stream: str, nats_category: str, job_type: str, job_n: int, embedding_model: str, generation_model: str)[source]

Bases: object

embedding_model: str

generation_model: str

job_n: int

job_type: str

milvus_collection: str

milvus_host: str

milvus_password: str

milvus_port: int

milvus_username: str

nats_category: str

nats_host: str

nats_password: str

nats_stream: str

nats_username: str

class magnet.utils.data_classes.MagnetConfig(host: str, domain: str = None, credentials: str = None, session: str = None, stream_name: str = None, category: str = None, kv_name: str = None, os_name: str = None, index: magnet.utils.data_classes.IndexConfig | None = None)[source]

Bases: object

category: str = None

credentials: str = None

domain: str = None

host: str

index: IndexConfig | None = None

kv_name: str = None

os_name: str = None

session: str = None

stream_name: str = None

class magnet.utils.data_classes.MistralArgs(dim: int, n_layers: int, head_dim: int, hidden_dim: int, n_heads: int, n_kv_heads: int, norm_eps: float, vocab_size: int)[source]

Bases: object

Represents a set of arguments for the Mistral model.

Args:: dim (int): The dimensionality of the model. n_layers (int): The number of layers in the model. head_dim (int): The dimensionality of each attention head. hidden_dim (int): The dimensionality of the hidden layer in the feed-forward network. n_heads (int): The number of attention heads. n_kv_heads (int): The number of attention heads used for key-value attention. norm_eps (float): The epsilon value used for numerical stability in layer normalization. vocab_size (int): The size of the vocabulary used in the model.

dim: int

head_dim: int

hidden_dim: int

n_heads: int

n_kv_heads: int

n_layers: int

norm_eps: float

vocab_size: int

class magnet.utils.data_classes.Payload(text: str, document: str)[source]

Bases: object

Represents a payload with two main fields: text and document.

Args:: text (str): The text associated with the payload. document (str): The document associated with the payload.

document: str

text: str

magnet.utils.globals module

class magnet.utils.globals.Utils[source]

Bases: object

The Utils class provides various utility functions for tasks such as checking CUDA availability, normalizing text, and uploading files to Amazon S3.

Example Usage:

# Create an instance of the Utils class utils = Utils()

# Check if CUDA is available utils.check_cuda()

# Normalize a text string normalized_text = utils.normalize_text(” This is a sample text. “)

# Upload a file to Amazon S3 utils.upload_to_s3(file_or_dir=”path/to/file.txt”, keys=(“YOUR_ACCESS_KEY”, “YOUR_SECRET_KEY”), bucket=”my-bucket”, bucket_path=”data”)

Main functionalities: - Checking if CUDA is available on the machine - Normalizing text by performing various cleaning operations - Uploading files to Amazon S3

Methods: - __init__(): Initializes the Utils class and sets up the spaCy English language model with a sentence tokenizer. - check_cuda(): Checks if CUDA is available on the machine and provides additional information if it is. - normalize_text(_): Cleans a text string by removing whitespace, replacing characters, and removing curly braces. - upload_to_s3(file_or_dir, keys, bucket, bucket_path): Uploads a file or directory to an Amazon S3 bucket using the provided access keys and bucket information.

Fields: The Utils class does not have any fields.

check_cuda()[source]: The function checks if CUDA is available on the machine and provides additional information if it is.

normalize_text(_)[source]

The clean function takes a string as input and performs various cleaning operations on it, such as removing whitespace, replacing characters, and removing curly braces.

Parameters:: _ – The parameter “_” is a placeholder for the input string that needs to be cleaned
Returns:: a cleaned version of the input string.

upload_to_s3(file_or_dir: str = None, keys: tuple = ('YOUR_ACCESS_KEY', 'YOUR_SECRET_KEY'), bucket: str = None, bucket_path: str = None)[source]

Uploads a file or directory to an Amazon S3 bucket.

Args:

file_or_dir (str): The path of the file or directory to be uploaded. keys (tuple): A tuple containing the access keys (access key ID and secret access key) required to access the Amazon S3 bucket. Default is (“YOUR_ACCESS_KEY”, “YOUR_SECRET_KEY”). bucket (str): The name of the Amazon S3 bucket where the file or directory will be uploaded. bucket_path (str): The path within the bucket where the file or directory will be uploaded.

Returns:

None

Raises:

None

Example Usage:

# Create an instance of the Utils class utils = Utils()

# Upload a file to Amazon S3 utils.upload_to_s3(file_or_dir=”path/to/file.txt”, keys=(“YOUR_ACCESS_KEY”, “YOUR_SECRET_KEY”), bucket=”my-bucket”, bucket_path=”data”)

magnet.utils.globals.break_into_chunks(text, max_words)[source]

Break a text into chunks of a specified maximum number of words.

Parameters: - text (str): The text to break into chunks. - max_words (int): The maximum number of words per chunk.

magnet.utils package

Subpackages

Submodules

magnet.utils.data_classes module

magnet.utils.globals module

magnet.utils.huggingface module

magnet.utils.local module

magnet.utils.markdown module

magnet.utils.milvus module

magnet.utils.prompts module