--- title: Setting up Hugging Face Account date: 2026-05-28 source: "Section 6 / Lecture 2" type: lecture-notes status: in-progress section: "Section 6: Running LLMs via Hugging Face Hub" tags: - hugging-face - account-setup - signup - foundations - hands-on related: - "[[01 - Intro to Hugging Face]]" - "[[03 - Accessing Gated Models]]" --- # Setting up Hugging Face Account > [!NOTE] > **TL;DR** > Quick setup chore — sign up at **huggingface.co**, solve the captcha, choose username + full name, confirm via email. Account is free for all basic use (browsing models, downloading open weights, light Spaces usage). Paid tiers exist for **PRO** (better Spaces, faster inference) and **Enterprise** (compliance, private models). A free account is sufficient for everything in this section. After login, the account unlocks: pulling models that require authentication ([[03 - Accessing Gated Models]]), generating API tokens ([[04 - Hugging Face CLI Setup and Login]]), pushing models, creating Spaces. > [!NOTE] > **Where this fits** > Second note of **Section 6: Running LLMs via Hugging Face Hub**. Pure account-setup chore. The interesting work starts in [[03 - Accessing Gated Models]]. --- ## 1. Why an account is needed at all Most actions on Hugging Face **can** be done without an account: - Browsing models, datasets, spaces. - Reading docs. - Downloading **fully open** models. - Using public Spaces (limited). But many useful actions require login: | Action | Account required? | |---|---| | Browse | ❌ | | Download fully-open models | ❌ | | **Download gated models** (Gemma, Llama 3, etc.) | ✅ | | **Generate API tokens** (for CLI / `transformers`) | ✅ | | **Push** models / datasets / Spaces | ✅ | | **Comment** in discussions | ✅ | | **Run higher-tier Spaces** | ✅ (often paid) | | **Inference Endpoints** (managed API hosting) | ✅ (paid) | The main reason here is the **gated models** (next note) — Google's Gemma, Meta's Llama, and others require license acceptance, which needs a logged-in account. --- ## 2. The sign-up flow ### Step 1 — Visit huggingface.co Click **Sign Up** in the top right. ### Step 2 — Captcha A bot-detection puzzle ("select all images with hats" or similar). Standard reCAPTCHA-style flow — click through until it's satisfied. ### Step 3 — Email + password Pick an email account that gets checked (the confirmation email arrives there). Set a strong password. ### Step 4 — Username + full name - **Username**: globally unique, becomes part of any model/space URL (e.g., `huggingface.co/jayanth/my-model`). Choose carefully — changes later are painful. - **Full name**: shows on the profile. Can be a pseudonym. - Other fields (bio, company, etc.) are optional. ### Step 5 — Optional avatar HF includes an **AI-generated avatar** generator. Type a prompt (e.g. *"tech octopus wearing specs"*) → a unique avatar is produced. Or upload a custom image. Both fine. ### Step 6 — Accept terms → submit ### Step 7 — Email confirmation HF sends a confirmation link. Click it. Account is now active and verified. --- ## 3. Suggested settings to configure right away After login, two settings panels are worth visiting once: | Setting | Why | |---|---| | **Profile → Edit profile** | Add a one-line bio. Makes the account look real, not abandoned. | | **Settings → Notifications** | Default notification volume is high — adjust if email gets noisy. | | **Settings → Access Tokens** | Needed in [[04 - Hugging Face CLI Setup and Login]]. Worth bookmarking the page. | | **Settings → SSH Keys** | For pushing repositories via SSH (not needed for this course). | | **Settings → Billing** | Confirms free tier; only relevant if upgrading. | --- ## 4. Free vs paid tiers | Tier | Cost | Includes | |---|---|---| | **Free** | $0 | Browse, download, push models/Spaces, basic Inference API | | **PRO** | ~$9/mo | Higher Inference API rate limits, ZeroGPU access for Spaces, early features | | **Enterprise Hub** | Custom | Private model registries, compliance, audit logs, SSO | | **Inference Endpoints** | Pay-per-use | Managed model serving on dedicated hardware | | **Spaces with GPU** | ~$0.40 – $4 per hour | Higher hardware tiers for Spaces | For everything in this section → **Free tier is enough**. --- ## 5. Username choice — practical advice Since the username becomes a permanent part of any URL: - Use the same handle as GitHub / Twitter when possible (recognizability). - Avoid version numbers or dates (`-2024`, `-v2`) — they age badly. - Keep it short — URLs include it: `huggingface.co//`. - Avoid underscores — hyphens are more URL-friendly. - Don't use temp emails for the account — if recovery is needed later, a real email matters. If a really cool username is already taken, claim variations early (the username "namesquatting" problem is real on HF too). --- ## 6. Multi-factor authentication (recommended) HF supports **TOTP-based 2FA** (Settings → Account → Two-factor authentication). Worth enabling, because: - The account gains value over time (custom models pushed, tokens generated). - Stolen access tokens can be used to read private repos. - A 2FA prompt is a 5-second cost for a meaningful security boost. --- ## 7. State after this step | Component | Status | |---|---| | HF account | ✅ Created | | Email confirmed | ✅ | | Profile basics | ✅ | | Access tokens | ❌ (Note 04) | | Gated model access | ❌ (Note 03) | | Local CLI / library | ❌ (Notes 04 / 05) | --- ## 8. Common gotchas > [!WARNING] > **First-time sign-up issues** | Symptom | Cause | Fix | |---|---|---| | Captcha keeps failing | Borderline image classification | Try again; switch network if VPN is interfering | | Confirmation email not arriving | Spam folder / wrong email typed | Check spam, or trigger resend | | Username taken | Common name | Add a small unique suffix | | Forgot password | Common | "Forgot password" link works fine | | Region-restricted access | Some countries have restrictions on certain models | Use VPN if blocked, or skip those models | | Login loop after MFA | Browser cookies | Clear cookies, retry | --- ## 9. Main takeaways - Sign-up at **huggingface.co** is quick — captcha, email, username, confirm. - **Free tier** covers everything in this section. - Username becomes part of all repo URLs → choose deliberately. - Account is needed for **gated models**, **API tokens**, and pushing artifacts. - Worth enabling **2FA** right away. - After this, the account can: - Accept license terms for gated models ([[03 - Accessing Gated Models]]). - Generate access tokens for the CLI ([[04 - Hugging Face CLI Setup and Login]]). - Be used by `transformers` to download authenticated models ([[05 - Using the Transformers Package]]). --- ## 10. Things to dig into - **HF settings docs**: https://huggingface.co/docs/hub/account-management - **Access tokens guide**: https://huggingface.co/docs/hub/security-tokens - **HF organizations**: free for teams; useful if collaborating. --- ## 11. Next up - [ ] [[03 - Accessing Gated Models]] — get permission to use Google's Gemma 3. --- ## Related - [[01 - Intro to Hugging Face]] — what HF is. - [[03 - Accessing Gated Models]] — first thing to do with the new account. ## Sources - **HF account management docs**: https://huggingface.co/docs/hub/account-management - **HF access tokens guide**: https://huggingface.co/docs/hub/security-tokens - **HF pricing**: https://huggingface.co/pricing