llama.cpp/examples/infill
vvhg1 11ea5c7d96
infill. : fix tokenization (#3508)
* infill tokens correction

* serverinfill tokens correction

* removing any leading whitespace from infill suffix and removing leeading space token from suffix when params.escape

* removing any leading whitespace from infill suffix and removing leeading space token from suffix when params.escape

* only rm when params.escape, rm space if possible which is added back or rm added space token

* only rm when params.escape, rm space if possible which is added back or rm added space token

* Revert "only rm when params.escape, rm space if possible which is added back or rm added space token"

This reverts commit 63ba0b621f.

* fix interactive prompt escaping and fix server infill leading space handling

* rm unnecessary bool check
2023-10-10 10:31:21 +03:00
..
CMakeLists.txt infill : add new example + extend server API (#3296) 2023-10-02 10:42:02 +03:00
infill.cpp infill. : fix tokenization (#3508) 2023-10-10 10:31:21 +03:00
README.md infill : add new example + extend server API (#3296) 2023-10-02 10:42:02 +03:00

llama.cpp/example/infill

This example shows how to use the infill mode with Code Llama models supporting infill mode. Currently the 7B and 13B models support infill mode.

Infill supports most of the options available in the main example.

For further information have a look at the main README.md in llama.cpp/example/main/README.md

Common Options

In this section, we cover the most commonly used options for running the infill program with the LLaMA models:

  • -m FNAME, --model FNAME: Specify the path to the LLaMA model file (e.g., models/7B/ggml-model.bin).
  • -i, --interactive: Run the program in interactive mode, allowing you to provide input directly and receive real-time responses.
  • -n N, --n-predict N: Set the number of tokens to predict when generating text. Adjusting this value can influence the length of the generated text.
  • -c N, --ctx-size N: Set the size of the prompt context. The default is 512, but LLaMA models were built with a context of 2048, which will provide better results for longer input/inference.

Input Prompts

The infill program provides several ways to interact with the LLaMA models using input prompts:

  • --in-prefix PROMPT_BEFORE_CURSOR: Provide the prefix directly as a command-line option.
  • --in-suffix PROMPT_AFTER_CURSOR: Provide the suffix directly as a command-line option.
  • --interactive-first: Run the program in interactive mode and wait for input right away. (More on this below.)

Interaction

The infill program offers a seamless way to interact with LLaMA models, allowing users to receive real-time infill suggestions. The interactive mode can be triggered using --interactive, and --interactive-first

Interaction Options

  • -i, --interactive: Run the program in interactive mode, allowing users to get real time code suggestions from model.
  • --interactive-first: Run the program in interactive mode and immediately wait for user input before starting the text generation.
  • --color: Enable colorized output to differentiate visually distinguishing between prompts, user input, and generated text.

Example

./infill -t 10 -ngl 0 -m models/codellama-13b.Q5_K_S.gguf -c 4096 --temp 0.7 --repeat_penalty 1.1 -n 20 --in-prefix "def helloworld():\n    print(\"hell" --in-suffix "\n   print(\"goodbye world\")\n    "