Chai Scraper
Spider read chai.ml in 145 ms without a browser and returned 149 lines of clean markdown, including sections like "1.4 EXAFLOP CLUSTER", "Larger Model Upgrade - 13B Architecture" and "PPO Model Deployed".
# 1.4 EXAFLOP CLUSTERWe continued to iterate on RLHF (Reinforcement Learning with Human Feedback), training a reward model directly on user signals. This led to a huge boost in our day 30 user retention.### Larger Model Upgrade - 13B ArchitectureWe found that a bigger model leads to better depth, therefore better retention. We re-trained our LLM from scratch and saw another +10% engagement boost.### PPO Model DeployedUsing Proximal Policy Optimization, a reinforcement learning technique, we optimized our base foundation model to decrease the probability a chat session ends.### Deploys Reward Model XLContinued to scale up our reward model. Trained with 100 million signals to decrease user retry rate and increase chat session length.### Efficient Inference & Custom GPU OrchestrationOff-the-shelf load balancing and vLLM were no longer sufficient to support our user base at 500K DAU scale. We implemented custom CUDA kernels together with our own GPU orchestration system.### Increased GPU ReservationWe hit a scaling issue due to high demand from our users. We reserved an additional 1,000 A100 GPUs from our provider to scale reliably.### Deployed Model BlendingCHAI invented model blending—ensembling different LLMs trained on different targets at the conversation level. This outperformed GPT-3 in user retention.### BO8 Reward Model DeployedWith increased cluster capacity, we implemented Best-of-8 rejection sampling, utilizing our upgraded reward model to its full extent.### DPO Model DeployedUtilizing Direct Preference Optimization with user preference datasets, we boosted engagement by 20%. The performance stacked well with our existing reward model.### Upgraded All Existing Blends to DPOBuilding on the success of DPO, we iterated on optimization targets and data selection, and successfully deployed DPO across all production blends.### 13B Reward Model DeployedThe same call, in code.
The capture above came back as markdown. These examples add a key, so you get browser rendering, proxies, and concurrency on chai.ml.
import { SpiderBrowser } from "spider-browser";
const spider = new SpiderBrowser({
apiKey: process.env.SPIDER_API_KEY!,
});
await spider.connect();
const page = spider.page!;
await page.goto("https://chai.ml");
// No selectors, no schema. Spider reads the page and names the fields.
const data = await page.scrape();
console.log(data);
await spider.close();import { Spider } from "@spider-cloud/spider-client";
const spider = new Spider({ apiKey: process.env.SPIDER_API_KEY! });
const result = await spider.scrapeUrl("https://www.chai.ml", {
return_format: "markdown",
});
console.log(result);Ready for volume? Get an API key →
Fields you can pull.
Spider names these from the page. The capture above came back as markdown; the same
call with return_format: "json" returns them as keys.
What chai.ml costs to scrape.
The capture above cost $0.000125 to fetch. Pricing is $1 per GB of pre-transformation bandwidth plus $0.001 per CPU minute, so a page like this one lands at a fraction of a cent. Failed requests are billed at $0.
- Free balance on signup
- No card required to test
- Balance never expires
Run it keyless, no account
More International scrapers.
Mercado Libre Scraper
Extract product listings, seller ratings, pricing in local currencies, and shipping data from Mercado Libre.
Rakuten Scraper
Extract product listings, store ratings, cashback offers, and pricing data from Rakuten Japan marketplace.
Flipkart Scraper
Extract product listings, seller data, pricing in INR, and delivery estimates from Flipkart India store.
Start scraping chai.ml.
You already have the call. A key raises the rate limit and turns on browser rendering, proxies, and concurrency. Balance never expires, and top-ups go through secure checkout.