warning

🚧 Cortex.cpp is currently under development. Our documentation outlines the intended behavior of Cortex, which may not yet be fully implemented in the codebase.

cortex models

This command allows you to start, stop, and manage various local or remote model operations within Cortex.

Usage:

info

You can use the --verbose flag to display more detailed output of the internal processes. To apply this flag, use the following format: cortex --verbose [subcommand].

MacOs/Linux

# Stable

cortex models [options] [subcommand]

# Beta

cortex-beta models [options] [subcommand]

# Nightly

cortex-nightly models [options] [subcommand]

Windows

# Stable

cortex.exe models [options]

# Beta

cortex-beta.exe models [options]

# Nightly

cortex-nightly.exe models [options]

Options:

Option	Description	Required	Default value	Example
-h, --help	Display help information for the command.	No	-	-h

cortex models get

info

This CLI command calls the following API endpoint:

• Get Model

This command returns a model detail defined by a model_id.

Usage:

info

You can use the --verbose flag to display more detailed output of the internal processes. To apply this flag, use the following format: cortex --verbose [subcommand].

MacOs/Linux

# Stable

cortex models get <model_id>

# Beta

cortex-beta models get <model_id>

# Nightly

cortex-nightly models get <model_id>

Windows

# Stable

cortex.exe models get <model_id>

# Beta

cortex-beta.exe models get <model_id>

# Nightly

cortex-nightly.exe models get <model_id>

For example, it returns the following:

ModelConfig Details:

-------------------

id: tinyllama

name: tinyllama 1B

model: tinyllama:1B

version: 1

stop: [</s>]

top_p: 0.95

temperature: 0.7

frequency_penalty: 0

presence_penalty: 0

max_tokens: 4096

stream: true

ngl: 33

ctx_len: 4096

engine: llamacpp

prompt_template:

<|system|>

{system_message}</s>

<|user|>

{prompt}</s>

<|assistant|>

system_template:

<|system|>

user_template: </s>

<|user|>

ai_template: </s>

<|assistant|>

tp: 0

text_model: false

files: [model_path]

created: 1725342964

info

This command uses a model_id from the model that you have downloaded or available in your file system.

Options:

Option	Description	Required	Default value	Example
model_id	The identifier of the model you want to retrieve.	Yes	-	mistral
-h, --help	Display help information for the command.	No	-	-h

cortex models list

info

This CLI command calls the following API endpoint:

• List Model

This command lists all the downloaded local and remote models.

Usage:

info

You can use the --verbose flag to display more detailed output of the internal processes. To apply this flag, use the following format: cortex --verbose [subcommand].

MacOs/Linux

# Stable

cortex models list [options]

# Beta

cortex-beta models list [options]

# Nightly

cortex-nightly models list [options]

Windows

# Stable

cortex.exe models list [options]

# Beta

cortex-beta.exe models list [options]

# Nightly

cortex-nightly.exe models list [options]

For example, it returns the following:

+---------+----------------+-----------------+---------+

| (Index) | ID | engine | version |

+---------+----------------+-----------------+---------+

| 1 | tinyllama-gguf | llamacpp | 1 |

+---------+----------------+-----------------+---------+

| 2 | tinyllama | llamacpp | 1 |

+---------+----------------+-----------------+---------+

Options:

Option	Description	Required	Default value	Example
-h, --help	Display help for command.	No	-	-h

cortex models start

info

This CLI command calls the following API endpoint:

• Start Model

This command starts a model defined by a model_id.

Usage:

info

You can use the --verbose flag to display more detailed output of the internal processes. To apply this flag, use the following format: cortex --verbose [subcommand].

MacOs/Linux

# Stable

cortex models start [options] <model_id>

# Beta

cortex-beta models start [options] <model_id>

# Nightly

cortex-nightly models start [options] <model_id>

Windows

# Stable

cortex.exe models start [options] <model_id>

# Beta

cortex-beta.exe models start [options] <model_id>

# Nightly

cortex-nightly.exe models start [options] <model_id>

info

This command uses a model_id from the model that you have downloaded or available in your file system.

Options:

Option	Description	Required	Default value	Example
model_id	The identifier of the model you want to start.	Yes	Prompt to select from the available models	mistral
-h, --help	Display help information for the command.	No	-	-h

cortex models stop

info

This CLI command calls the following API endpoint:

• Stop Model

This command stops a model defined by a model_id.

Usage:

info

You can use the --verbose flag to display more detailed output of the internal processes. To apply this flag, use the following format: cortex --verbose [subcommand].

MacOs/Linux

# Stable

cortex models stop <model_id>

# Beta

cortex-beta models stop <model_id>

# Nightly

cortex-nightly models stop <model_id>

Windows

# Stable

cortex.exe models stop <model_id>

# Beta

cortex-beta.exe models stop <model_id>

# Nightly

cortex-nightly.exe models stop <model_id>

info

This command uses a model_id from the model that you have started before.

Options:

Option	Description	Required	Default value	Example
model_id	The identifier of the model you want to stop.	Yes	-	mistral
-h, --help	Display help information for the command.	No	-	-h

cortex models update

info

This CLI command calls the following API endpoint:

• Update Model

This command updates a model configuration defined by a model_id.

Usage:

info

You can use the --verbose flag to display more detailed output of the internal processes. To apply this flag, use the following format: cortex --verbose [subcommand].

MacOs/Linux

# Stable

cortex models update [options] <model_id>

# Beta

cortex-beta models update [options] <model_id>

# Nightly

cortex-nightly models update [options] <model_id>

Windows

# Stable

cortex.exe models update [options] <model_id>

# Beta

cortex-beta.exe models update [options] <model_id>

# Nightly

cortex-nightly.exe models update [options] <model_id>

info

This command uses a model_id from the model that you have downloaded or available in your file system.

Options:

Option	Description	Required	Default value	Example
model_id	The identifier of the model you want to update.	Yes	-	mistral
-c, --options <options...>	Specify the options to update the model. Syntax: -c option1=value1 option2=value2.	Yes	-	-c max_tokens=100 temperature=0.5
-h, --help	Display help information for the command.	No	-	-h

cortex models delete

info

This CLI command calls the following API endpoint:

• Delete Model

This command deletes a local model defined by a model_id.

Usage:

info

You can use the --verbose flag to display more detailed output of the internal processes. To apply this flag, use the following format: cortex --verbose [subcommand].

MacOs/Linux

# Stable

cortex models delete <model_id>

# Beta

cortex-beta models delete <model_id>

# Nightly

cortex-nightly models delete <model_id>

Windows

# Stable

cortex.exe models delete <model_id>

# Beta

cortex-beta.exe models delete <model_id>

# Nightly

cortex-nightly.exe models delete <model_id>

info

This command uses a model_id from the model that you have downloaded or available in your file system.

Options:

Option	Description	Required	Default value	Example
model_id	The identifier of the model you want to delete.	Yes	-	mistral
-h, --help	Display help for command.	No	-	-h

cortex models alias

This command adds an alias to a local model that function the same as model_id.

Usage:

info

You can use the --verbose flag to display more detailed output of the internal processes. To apply this flag, use the following format: cortex --verbose [subcommand].

MacOs/Linux

# Stable

cortex models alias --model_id <model_id> --alias <new_model_id_or_model_alias>

# Beta

cortex-beta models alias --model_id <model_id> --alias <new_model_id_or_model_alias>

# Nightly

cortex-nightly models alias --model_id <model_id> --alias <new_model_id_or_model_alias>

Windows

# Stable

cortex.exe models alias --model_id <model_id> --alias <new_model_id_or_model_alias>

# Beta

cortex-beta.exe models alias --model_id <model_id> --alias <new_model_id_or_model_alias>

# Nightly

cortex-nightly.exe models alias --model_id <model_id> --alias <new_model_id_or_model_alias>

Options:

Option	Description	Required	Default value	Example
--model_id	The identifier of the model.	Yes	-	mistral
-alias	The new identifier for the model.	Yes	-	mistral_2

cortex models update

This command updates the model.yaml file of a local model.

Usage:

info

You can use the --verbose flag to display more detailed output of the internal processes. To apply this flag, use the following format: cortex --verbose [subcommand].

MacOs/Linux

# Stable

cortex models update [options]

# Beta

cortex-beta models update [options]

# Nightly

cortex-nightly models update [options]

Windows

# Stable

cortex.exe models update [options]

# Beta

cortex-beta.exe models update [options]

# Nightly

cortex-nightly.exe models update [options]

Options:

Option	Description	Required	Default value	Example
-h, --help	Display help for command.	No	-	-h
--model_id REQUIRED	Unique identifier for the model.	Yes	-	--model_id my_model
--name	Name of the model.	No	-	--name "GPT Model"
--model	Model type or architecture.	No	-	--model GPT-4
--version	Version of the model to use.	No	-	--version 1.2.0
--stop	Stop token to terminate generation.	No	-	--stop "</s>"
--top_p	Sampling parameter for nucleus sampling.	No	-	--top_p 0.9
--temperature	Controls randomness in generation.	No	-	--temperature 0.8
--frequency_penalty	Penalizes repeated tokens based on frequency.	No	-	--frequency_penalty 0.5
--presence_penalty	Penalizes repeated tokens based on presence.	No	0.0	--presence_penalty 0.6
--max_tokens	Maximum number of tokens to generate.	No	-	--max_tokens 1500
--stream	Stream output tokens as they are generated.	No	false	--stream true
--ngl	Number of generations in parallel.	No	-	--ngl 4
--ctx_len	Maximum context length in tokens.	No	-	--ctx_len 1024
--engine	Compute engine for running the model.	No	-	--engine CUDA
--prompt_template	Template for the prompt structure.	No	-	--prompt_template "###"
--system_template	Template for system-level instructions.	No	-	--system_template "SYSTEM"
--user_template	Template for user inputs.	No	-	--user_template "USER"
--ai_template	Template for AI responses.	No	-	--ai_template "ASSISTANT"
--os	Operating system environment.	No	-	--os Ubuntu
--gpu_arch	GPU architecture specification.	No	-	--gpu_arch A100
--quantization_method	Quantization method for model weights.	No	-	--quantization_method int8
--precision	Floating point precision for computations.	No	float32	--precision float16
--tp	Tensor parallelism.	No	-	--tp 4
--trtllm_version	Version of the TRTLLM library.	No	-	--trtllm_version 2.0
--text_model	The model used for text generation.	No	-	--text_model llama2
--files	File path or resources associated with the model.	No	-	--files config.json
--created	Creation date of the model.	No	-	--created 2024-01-01
--object	The object type (e.g., model or file).	No	-	--object model
--owned_by	The owner or creator of the model.	No	-	--owned_by "Company"
--seed	Seed for random number generation.	No	-	--seed 42
--dynatemp_range	Range for dynamic temperature scaling.	No	-	--dynatemp_range 0.7-1.0
--dynatemp_exponent	Exponent for dynamic temperature scaling.	No	-	--dynatemp_exponent 1.2
--top_k	Top K sampling to limit token selection.	No	-	--top_k 50
--min_p	Minimum probability threshold for tokens.	No	-	--min_p 0.1
--tfs_z	Token frequency selection scaling factor.	No	-	--tfs_z 0.5
--typ_p	Typicality-based token selection probability.	No	-	--typ_p 0.9
--repeat_last_n	Number of last tokens to consider for repetition penalty.	No	-	--repeat_last_n 64
--repeat_penalty	Penalty for repeating tokens.	No	-	--repeat_penalty 1.2
--mirostat	Mirostat sampling method for stable generation.	No	-	--mirostat 1
--mirostat_tau	Target entropy for Mirostat.	No	-	--mirostat_tau 5.0
--mirostat_eta	Learning rate for Mirostat.	No	-	--mirostat_eta 0.1
--penalize_nl	Penalize new lines in generation.	No	false	--penalize_nl true
--ignore_eos	Ignore the end of sequence token.	No	false	--ignore_eos true
--n_probs	Number of probability outputs to return.	No	-	--n_probs 5

cortex models import

This command imports the local model using the model's gguf file.

Usage:

info

You can use the --verbose flag to display more detailed output of the internal processes. To apply this flag, use the following format: cortex --verbose [subcommand].

MacOs/Linux

# Stable

cortex models import --model_id <model_id> --model_path </path/to/your/model.gguf>

# Beta

cortex-beta models import --model_id <model_id> --model_path </path/to/your/model.gguf>

# Nightly

cortex-nightly models import --model_id <model_id> --model_path </path/to/your/model.gguf>

Windows

# Stable

cortex.exe models import --model_id <model_id> --model_path </path/to/your/model.gguf>

# Beta

cortex-beta.exe models import --model_id <model_id> --model_path </path/to/your/model.gguf>

# Nightly

cortex-nightly.exe models import --model_id <model_id> --model_path </path/to/your/model.gguf>

Options:

Option	Description	Required	Default value	Example
-h, --help	Display help for command.	No	-	-h
--model_id	The identifier of the model.	Yes	-	mistral
--model_path	The path of the model source file.	Yes	-	/path/to/your/model.gguf