AWS bedrock documentation change
Summary
Updated reinforcement fine-tuning job documentation with corrected encryption link, adjusted UI workflow steps, parameter values (inferenceMaxTokens/maxPromptLength swapped, learningRate/trainingSamplePerPrompt changed), and API configuration fixes.
Security assessment
The encryption reference change only adds a hyperlink to existing documentation. Parameter adjustments and workflow changes appear to be usability/accuracy improvements rather than security fixes. No evidence of addressing vulnerabilities or security incidents.
Diff
diff --git a/bedrock/latest/userguide/rft-submit-job.md b/bedrock/latest/userguide/rft-submit-job.md index 18da4301e..fb2cdd4b0 100644 --- a//bedrock/latest/userguide/rft-submit-job.md +++ b//bedrock/latest/userguide/rft-submit-job.md @@ -32 +32 @@ You can create a reinforcement fine-tuning job using the Amazon Bedrock console - * (Optional) Encrypt input and output data, your RFT job, or inference requests made to custom models. For more information, see Encryption of custom models link. + * (Optional) Encrypt input and output data, your RFT job, or inference requests made to custom models. For more information, see [ Encryption of custom models](https://docs.aws.amazon.com/bedrock/latest/userguide/encryption-custom-job.html). @@ -50 +50 @@ To submit an RFT job in the console, carry out the following steps: - 3. In the Models table, choose **Customize model** and then **Create Reinforcement Fine-tuning job**. + 3. In the Models table, choose **Create**. Then, choose **Create reinforcement fine-tuning job**. @@ -52 +52 @@ To submit an RFT job in the console, carry out the following steps: - 4. In the **Model details** section, do the following: + 4. In the **Model details** section, choose **Amazon Nova 2 Lite** as your base model. @@ -54,5 +54 @@ To submit an RFT job in the console, carry out the following steps: - 1. Choose **Amazon Nova 2 Lite** as your base model - - 2. (Optional) Select **Model encryption** to use a custom KMS key - - 5. In the **Job configuration** section, enter a name for the job and optionally add tags. + 5. In the **Customization details** section, enter the customization name. @@ -76 +72 @@ Your training dataset should be in the OpenAI Chat Completions data format. If y -The console's "Model as judge" option automatically converts your configuration into a Lambda function during training. +The console's **Model as judge** option automatically converts your configuration into a Lambda function during training. @@ -86 +82 @@ For more information, see [Setting up reward functions](./reward-functions.html) - 10. In the **Service access** section, select: + 10. In the **Role configuration** section, select: @@ -100 +96 @@ For more information, see [Setting up reward functions](./reward-functions.html) - 12. Choose **Create Reinforcement Fine-tuning job** to begin the job. + 12. Choose **Create reinforcement fine-tuning job** to begin the job. @@ -122 +118 @@ Send a CreateModelCustomizationJob request with `customizationType` set to `REIN - * `trainingDataConfig` \- Amazon S3 URI of training dataset or API invocation log configuration + * `trainingDataConfig` \- Amazon S3 URI of training dataset or invocation log configuration @@ -126 +122 @@ Send a CreateModelCustomizationJob request with `customizationType` set to `REIN - * `rewardFunctionConfig` \- Reward function configuration (RLVR or RLAIF) + * `rftConfig` \- Reward function configuration (RLVR or RLAIF) and hyper paramerters configuration @@ -154,3 +150,3 @@ Send a CreateModelCustomizationJob request with `customizationType` set to `REIN - "inferenceMaxTokens": 4096, - "learningRate": learning-rate-value, - "maxPromptLength": 8192, + "inferenceMaxTokens": 8192, + "learningRate": 0.00001, + "maxPromptLength": 4096, @@ -158 +154 @@ Send a CreateModelCustomizationJob request with `customizationType` set to `REIN - "trainingSamplePerPrompt": 8 + "trainingSamplePerPrompt": 4 @@ -175 +171 @@ Send a CreateModelCustomizationJob request with `customizationType` set to `REIN - customizationType = "FINE_TUNING" + customizationType = "REINFORCEMENT_FINE_TUNING" @@ -192,5 +188,5 @@ Send a CreateModelCustomizationJob request with `customizationType` set to `REIN - 'inferenceMaxTokens': 4096, - 'learningRate': learning-rate-value, - 'maxPromptLength': 8192, - 'reasoningEffort': 'low', - 'trainingSamplePerPrompt': 8 + 'inferenceMaxTokens': 8192, + 'learningRate':0.00001, + 'maxPromptLength': 4096, + 'reasoningEffort': 'high', + 'trainingSamplePerPrompt':4 @@ -212 +208,2 @@ Send a CreateModelCustomizationJob request with `customizationType` set to `REIN - outputDataConfig=outputDataConfig + outputDataConfig=outputDataConfig, + customizationType=customizationType @@ -215 +212 @@ Send a CreateModelCustomizationJob request with `customizationType` set to `REIN - jobArn = response_ft.get('jobArn') + jobArn = response_ft['jobArn']