Create Fine-Tune Job with Lamini API on New Item in Feed from RSS API

Pipedream makes it easy to connect APIs for Lamini, RSS and 2,700+ other apps remarkably fast.

Trigger workflow on

New Item in Feed from the RSS API

Next, do this

Create Fine-Tune Job with the Lamini API

No credit card required

▶

Watch us build a workflow

8 min

Watch now ➜

Trusted by 1,000,000+ developers from startups to Fortune 500 companies

Developers ♥ Pipedream

Getting Started#

This integration creates a workflow with a RSS trigger and Lamini action. When you configure and deploy the workflow, it will run on Pipedream's servers 24x7 for free.

Select this integration
Configure the New Item in Feed trigger
1. Connect your RSS account
2. Configure timer
3. Configure Feed URL
4. Configure Published After
Configure the Create Fine-Tune Job action
1. Connect your Lamini account
2. Select a Model Name
3. Configure Dataset ID
4. Optional- Configure Finetune Arguments
5. Optional- Configure GPU Config
6. Optional- Configure Is Public
7. Optional- Configure Custom Model Name
8. Optional- Configure Wait for Completion
Deploy the workflow
Send a test event to validate your setup
Turn on the trigger

Details#

This integration uses pre-built, source-available components from Pipedream's GitHub repo. These components are developed by Pipedream and the community, and verified and maintained by Pipedream.

To contribute an update to an existing component or create a new component, create a PR on GitHub. If you're new to Pipedream component development, you can start with quickstarts for trigger span and action development, and then review the component API reference.

Trigger#

New Item in Feed on RSS

Description:Emit new items from an RSS feed

Version:1.2.7

Key:rss-new-item-in-feed

View on GitHub

RSS Overview#

The RSS app allows users to automatically fetch and parse updates from web feeds. This functionality is pivotal for staying abreast of content changes or updates from websites, blogs, and news outlets that offer RSS feeds. With Pipedream, you can harness the RSS API to trigger workflows that enable a broad range of automations, like content aggregation, monitoring for specific keywords, notifications, and data synchronization across platforms.

Trigger Code#

import { defineSource } from "@pipedream/types";
import rss from "../../app/rss.app.mjs";
import rssCommon from "../common/common.mjs";
export default defineSource({
    ...rssCommon,
    key: "rss-new-item-in-feed",
    name: "New Item in Feed",
    description: "Emit new items from an RSS feed",
    version: "1.2.7",
    type: "source",
    dedupe: "unique",
    props: {
        ...rssCommon.props,
        url: {
            propDefinition: [
                rss,
                "url",
            ],
        },
        publishedAfter: {
            type: "string",
            label: "Published After",
            description: "Emit items published after the specified date in ISO 8601 format .e.g `2022-12-07T12:57:10+07:00`",
            default: new Date(Date.now() - 24 * 60 * 60 * 1000).toISOString(), // default to one day ago
        },
    },
    hooks: {
        async activate() {
            // Try to parse the feed one time to confirm we can fetch and parse.
            // The code will throw any errors to the user.
            await this.rss.fetchAndParseFeed(this.url);
        },
    },
    methods: {
        ...rssCommon.methods,
        generateMeta: function (item) {
            return {
                id: this.rss.itemKey(item),
                summary: item.title,
                ts: Date.now(),
            };
        },
    },
    async run() {
        const items = await this.rss.fetchAndParseFeed(this.url);
        for (const item of items.reverse()) {
            const publishedAfter = +new Date(this.publishedAfter);
            const ts = this.rss.itemTs(item);
            if (Number.isNaN(publishedAfter) || publishedAfter > ts) {
                continue;
            }
            const meta = this.generateMeta(item);
            this.$emit(item, meta);
        }
    },
});

Trigger Configuration#

This component may be configured based on the props defined in the component code. Pipedream automatically prompts for input values in the UI and CLI.

Label	Prop	Type	Description
RSS	`rss`	`app`	This component uses the RSS app.
	`timer`	`$.interface.timer`	How often you want to poll the feed for new items
Feed URL	`url`	`string`	Enter the URL for any public RSS feed
Published After	`publishedAfter`	`string`	Emit items published after the specified date in ISO 8601 format .e.g `2022-12-07T12:57:10+07:00`

Trigger Authentication#

The RSS API does not require authentication.

About RSS#

Real Simple Syndication

Action#

Create Fine-Tune Job on Lamini

Description:Create a fine-tuning job with a dataset. [See the documentation](https://docs.lamini.ai/api/).

Version:0.0.2

Key:lamini-create-fine-tune-job

View on GitHub

Action Code#

import app from "../../lamini.app.mjs";
import constants from "../../common/constants.mjs";
import utils from "../../common/utils.mjs";

export default {
  key: "lamini-create-fine-tune-job",
  name: "Create Fine-Tune Job",
  description: "Create a fine-tuning job with a dataset. [See the documentation](https://docs.lamini.ai/api/).",
  version: "0.0.2",
  type: "action",
  props: {
    app,
    modelName: {
      description: "Base model to be fine-tuned.",
      propDefinition: [
        app,
        "modelName",
        () => ({
          includeFineTunedModels: false,
        }),
      ],
    },
    datasetId: {
      type: "string",
      label: "Dataset ID",
      description: "Previously uploaded dataset to use for training. Please use the **Upload Dataset** action to upload a dataset.",
    },
    fineTuneArgs: {
      type: "object",
      label: "Finetune Arguments",
      description: "Optional hyperparameters for fine-tuning. Each property is optional:\n- `index_pq_m`: Number of subquantizers for PQ (eg. 8)\n- `index_max_size`: Maximum index size (eg. 65536)\n- `max_steps`: Maximum number of training steps (eg. 60)\n- `batch_size`: Training batch size (eg. 1)\n- `learning_rate`: Learning rate (eg. 0.0003)\n- `index_pq_nbits`: Number of bits per subquantizer (eg. 8)\n- `max_length`: Maximum sequence length (eg. 2048)\n- `index_ivf_nlist`: Number of IVF lists (eg. 2048)\n- `save_steps`: Steps between checkpoints (eg. 60)\n- `args_name`: Name for the argument set (eg. \"demo\")\n- `r_value`: R value for LoRA (eg. 32)\n- `index_hnsw_m`: Number of neighbors in HNSW (eg. 32)\n- `index_method`: Indexing method (eg. \"IndexIVFPQ\")\n- `optim`: Optimizer to use (eg. \"adafactor\")\n- `index_hnsw_efConstruction`: HNSW construction parameter (eg. 16)\n- `index_hnsw_efSearch`: HNSW search parameter (eg. 8)\n- `index_k`: Number of nearest neighbors (eg. 2)\n- `index_ivf_nprobe`: Number of IVF probes (eg. 48)\n- `eval_steps`: Steps between evaluations (eg. 30)\n[See the documentation](https://docs.lamini.ai/tuning/hyperparameters/#finetune_args).",
      optional: true,
    },
    gpuConfig: {
      type: "object",
      label: "GPU Config",
      description: "Optional GPU configuration for fine-tuning. [See the documentation](https://docs.lamini.ai/tuning/hyperparameters/#gpu_config).",
      optional: true,
    },
    isPublic: {
      type: "boolean",
      label: "Is Public",
      description: "Whether this fine-tuning job and dataset should be publicly accessible.",
      optional: true,
    },
    customModelName: {
      type: "string",
      label: "Custom Model Name",
      description: "A human-readable name for the fine-tuned model.",
      optional: true,
    },
    waitForCompletion: {
      type: "boolean",
      label: "Wait for Completion",
      description: "If set to `true`, the action will wait and poll until the fine-tuning job is `COMPLETED`. If is set to `false`, it will return immediately after creating the job.",
      default: false,
      optional: true,
    },
  },
  methods: {
    createFineTuneJob(args = {}) {
      return this.app.post({
        versionPath: constants.VERSION_PATH.V1,
        path: "/train",
        ...args,
      });
    },
  },
  async run({ $ }) {
    const {
      app,
      createFineTuneJob,
      modelName,
      datasetId,
      fineTuneArgs,
      gpuConfig,
      isPublic,
      customModelName,
      waitForCompletion,
    } = this;

    const MAX_RETRIES = 15;
    const DELAY = 1000 * 30; // 30 seconds
    const { run } = $.context;

    // First run: Create the fine-tune job
    if (run.runs === 1) {
      const { upload_base_path: uploadBasePath } =
        await app.getUploadBasePath({
          $,
        });

      await app.getExistingDataset({
        $,
        data: {
          dataset_id: datasetId,
          upload_base_path: uploadBasePath,
        },
      });

      const response = await createFineTuneJob({
        $,
        data: {
          model_name: modelName,
          dataset_id: datasetId,
          upload_file_path: `${uploadBasePath}/${datasetId}.jsonlines`,
          finetune_args: utils.parseJson(fineTuneArgs),
          gpu_config: utils.parseJson(gpuConfig),
          is_public: isPublic,
          custom_model_name: customModelName,
        },
      });

      $.export("$summary", `Successfully created a fine-tune job with ID \`${response.job_id}\`.`);

      // If user doesn't want to wait, return immediately
      if (!waitForCompletion) {
        return response;
      }

      // Store job_id for polling and start rerun
      $.flow.rerun(DELAY, {
        jobId: response.job_id,
      }, MAX_RETRIES);
      return response;
    }

    // Subsequent runs: Poll for job status
    if (run.runs > MAX_RETRIES) {
      throw new Error("Max retries exceeded - fine-tuning job may still be running");
    }

    const { jobId } = run.context;

    // Poll for status
    const statusResponse = await app.getJobStatus({
      $,
      jobId,
    });

    // If job is completed, return the final status
    if (statusResponse.status === "COMPLETED") {
      $.export("$summary", `Fine-tuning job \`${jobId}\` completed successfully.`);
      return statusResponse;
    }

    // If job failed, throw error
    if (statusResponse.status === "FAILED") {
      throw new Error(`Fine-tuning job \`${jobId}\` failed.`);
    }

    // Otherwise, continue polling
    $.flow.rerun(DELAY, {
      jobId,
    }, MAX_RETRIES);
    return {
      status: statusResponse.status,
      jobId,
      message: `Job is still running. Current status: ${statusResponse.status}`,
    };
  },
};

Action Configuration#

This component may be configured based on the props defined in the component code. Pipedream automatically prompts for input values in the UI.

Label	Prop	Type	Description
Lamini	`app`	`app`	This component uses the Lamini app.
Model Name	`modelName`	`string`	Select a value from the drop down menu.
Dataset ID	`datasetId`	`string`	Previously uploaded dataset to use for training. Please use the Upload Dataset action to upload a dataset.
Finetune Arguments	`fineTuneArgs`	`object`	Optional hyperparameters for fine-tuning. Each property is optional: `index_pq_m`: Number of subquantizers for PQ (eg. 8) `index_max_size`: Maximum index size (eg. 65536) `max_steps`: Maximum number of training steps (eg. 60) `batch_size`: Training batch size (eg. 1) `learning_rate`: Learning rate (eg. 0.0003) `index_pq_nbits`: Number of bits per subquantizer (eg. 8) `max_length`: Maximum sequence length (eg. 2048) `index_ivf_nlist`: Number of IVF lists (eg. 2048) `save_steps`: Steps between checkpoints (eg. 60) `args_name`: Name for the argument set (eg. "demo") `r_value`: R value for LoRA (eg. 32) `index_hnsw_m`: Number of neighbors in HNSW (eg. 32) `index_method`: Indexing method (eg. "IndexIVFPQ") `optim`: Optimizer to use (eg. "adafactor") `index_hnsw_efConstruction`: HNSW construction parameter (eg. 16) `index_hnsw_efSearch`: HNSW search parameter (eg. 8) `index_k`: Number of nearest neighbors (eg. 2) `index_ivf_nprobe`: Number of IVF probes (eg. 48) `eval_steps`: Steps between evaluations (eg. 30) See the documentation
GPU Config	`gpuConfig`	`object`	Optional GPU configuration for fine-tuning. See the documentation
Is Public	`isPublic`	`boolean`	Whether this fine-tuning job and dataset should be publicly accessible.
Custom Model Name	`customModelName`	`string`	A human-readable name for the fine-tuned model.
Wait for Completion	`waitForCompletion`	`boolean`	If set to `true`, the action will wait and poll until the fine-tuning job is `COMPLETED`. If is set to `false`, it will return immediately after creating the job.