GenAIModel
Params
name
A human readable description of the object.
provider
description to be done
model_name
description to be done
server
An instance of GPUServer.
nb_of_bits_per_parameter
generative ai model nb of bits per parameter from hypothesis in dimensionless.
llm_memory_factor
generative ai model ratio between gpu memory footprint and model size from ecologits in dimensionless.
gpu_latency_alpha
gpu latency per active parameter and output token from ecologits in second.
gpu_latency_beta
base gpu latency per output_token from ecologits in second.
bits_per_token
number of bits per token from hypothesis in dimensionless.
Backwards links
Calculated attributes
active_params
ExplainableQuantity in dimensionless, representing the open-mistral-7b from mistralai nb of active parameters from ecologits.
Example value: 7300000000.0 dimensionless
Depends directly on:
through the following calculations:
You can also visit the link to open-mistral-7b from mistralai nb of active parameters from Ecologits’s full calculation graph.
total_params
ExplainableQuantity in dimensionless, representing the open-mistral-7b from mistralai total nb of parameters from ecologits.
Example value: 7300000000.0 dimensionless
Depends directly on:
through the following calculations:
You can also visit the link to open-mistral-7b from mistralai total nb of parameters from Ecologits’s full calculation graph.
base_ram_consumption
ExplainableQuantity in gigabyte, representing the generative ai model base ram consumption.
Example value: 17.52 gigabyte
Depends directly on:
- Generative AI model ratio between GPU memory footprint and model size from Ecologits
- open-mistral-7b from mistralai total nb of parameters from Ecologits
- Generative AI model nb of bits per parameter from hypothesis
through the following calculations:
You can also visit the link to Generative AI model base RAM consumption’s full calculation graph.