diff --git a/README.md b/README.md index da80cda..8a29c19 100644 --- a/README.md +++ b/README.md @@ -1,21 +1,32 @@ -# testnet-doc -Documentation on node and validators +# TestNet Documentation +Documentation on training nodes and validators + + +## Table of Contents + +1. [Fed Ledger](#fed-ledger) +2. [Training Node](#training-node) + - [How to participate a LLM training task](#how-to-participate-a-llm-training-task) +3. [Validator](#validator) + - [How to participate a LLM validation task](#how-to-participate-a-llm-validation-task) + +. ## Fed Ledger ## Training Node ### How to participate a LLM training task -To interact with FLock's Fed Ledger service, you have to obtain API key from `train.flock.io` first. - -If you want to submit the model for a given training task, you have to stake through the Web UI for that task, or else Fed Ledger will not accept your submission. +To interact with FLock's Fed Ledger service, first obtain an API key from `train.flock.io`. -#### Get details of a task +**Staking Requirement**: To submit a model for a training task, you must stake through the Web UI. Otherwise, Fed Ledger will not accept your submission. +#### Getting Task Details +To get details of a specific task, use the following command: ```bash curl --location 'https://fed-ledger-staging.flock.io/api/v1/tasks/get?task_id=' ``` -#### Submit your trained model -To submit your trained model, you have to upload all model files to your public huggingface model repo first and then provide the repo name. +#### Submitting Your Trained Models +After training your model, you need to upload all model files to your public Hugging Face model repository. Then, provide the repo name with the following command: ```bash curl --location 'https://fed-ledger-staging.flock.io/api/v1/tasks/submit-result' \ @@ -31,20 +42,21 @@ curl --location 'https://fed-ledger-staging.flock.io/api/v1/tasks/submit-result' ## Validator -### How to participate as a LLM training task validator - -Similar to training task, you have to stake as a validator for a given task before you can request a task submission to validate on. +### How to participate a LLM validation task -#### Request a random task submission for a given task +Similar to the training task, to act as a validator, you must stake for a given task before you can request a submission to validate. +#### Requesting a Validation Assignment +To request a random task submission for validation, use the following command: ```bash curl --location --request POST 'https://fed-ledger-staging.flock.io/api/v1/tasks/request-validation-assignment/' \ --header 'flock-api-key: ' ``` -In response, you will get a `id` for this validation assignment you are assigned. You need to provide this `id` when you upload the validation result. +In response, you will get an `id` for this validation assignment you are assigned. You need to provide this `id` when you upload the validation result. -After you run the validation script and get the score, you can upload the validation result through +#### Submitting Validation Results +Once assigned a validation task, run the validation script to obtain the score, then submit the validation result with the following command: ```bash curl --location 'https://fed-ledger-staging.flock.io/api/v1/tasks/update-validation-assignment/' \ @@ -58,7 +70,7 @@ curl --location 'https://fed-ledger-staging.flock.io/api/v1/tasks/update-validat }' ``` -If you couldn't complete the validation task for this submission, you can manually submit a failed result to Fed Ledger so that you can continue on other validation assignments. +**Submitting a failed result**: If you're unable to complete the validation task, you can manually submit a failed result to Fed Ledger so that you can continue on other validation assignments, with this command: ```bash curl --location 'https://fed-ledger-staging.flock.io/api/v1/tasks/update-validation-assignment/' \