Program profile · official sources
Transformers in Practice
DeepLearning.AI and AMD offer a practical course on transformer behavior, internals and deployment. Three modules cover autoregressive generation, attention and model layers, then GPU inference and optimization. Visualizations, labs and quizzes address sampling, structured output, quantization, KV caching, flash attention and speculative decoding. Learners should understand basic neural network concepts and have experience using LLMs. Pro members can earn a course certificate; membership is a separate recurring access cost.
Facts checked 15 Oct 2026
- Price
- See pricing
As shown in the United States on 15 October 2026, the course FAQ quotes DeepLearning.AI Pro at USD 30 per month billed monthly, or USD 25 per month billed annually. These are recurring membership quotes, not a fixed completion cost; an annual charged total is not separately stated and is not calculated here. Taxes may apply by location. The certificate requires Pro.
- Provider link
See price and enroll at DeepLearning.AI (official site) ↗
Official site. No commission.
- Time
- 11h1m listed course duration; individual activity estimates are rounded
- Level
- Intermediate
- Credential
- Course certificate with Pro
- Prerequisites
- Learners should be comfortable using LLMs through an API or chat interface and understand basic neural network concepts such as weights, layers and training. Prior model-building from scratch is not required.
- Format
- Online course with video lessons and listed learning activities
Commercial status: no commercial relationship is documented for this page. The verification link goes directly to the official provider source.
Sources ↗
No fixed course or completion price is recorded in this registry; the dated Pro membership quotes are in the sourced pricing details above.
See the full registry price landscape. Prices retain their stated payment scope; they are not comparable completion costs.
What’s covered
- Module 1: Observed behavior
- Module 2: LLM internals and attention
- Module 3: Scaling and deploying
Prerequisites: Learners should be comfortable using LLMs through an API or chat interface and understand basic neural network concepts such as weights, layers and training. Prior model-building from scratch is not required.
What you actually do
- Module 1: Observed behavior : Conversation between Sharon Zhou and Andrew Ng Video ・ 4m Transformers in practice Video ・ 2m The autoregressive loop Video ・ 4m Visualization tutorial Video ・ 3m Visualization: The autoregressive Loop Code Example ・ 10m Token sampling Video ・ 5m Visualization: Selecting the Next Token Code Example ・ 10m Autoregressive dynamics Video ・ 3m Visualization: How Sampling Shapes Output Code Example ・ 10m Structured outputs Video ・ 5m Visualization: Constrained Generation with Finite State Machines (FSM) Code Example ・ 10m Grounding in context Video ・ 7m Thinking and reasoning Video ・ 5m Additional Readings for Module 1 Reading ・ 10m Module 1: Graded Lab Graded ・Code Assignment ・ 1h Module 1: Quiz Graded ・Quiz ・ 30m Join the DeepLearning.AI Forum to ask questions, get support, or share amazing ideas! Reading ・ 1m · Source
- Module 2: LLM internals and attention : Attention Video ・ 8m Visualization: Decoder-Only Transformers Code Example ・ 10m Visualization: Interpretable Attention Heads Code Example ・ 1h Positional encoding Video ・ 4m Model layers Video ・ 6m Decoding Video ・ 4m Visualization: Decoding Intermediate Layers Code Example ・ 1h Ungraded Lab Code Example ・ 10m Additional Readings for Module 2 Reading ・ 10m Module 2: Graded Lab Graded ・Code Assignment ・ 1h Module 2: Quiz Graded ・Quiz ・ 30m · Source
- Module 3: Scaling and deploying : Putting the model in production Video ・ 4m Quantization Video ・ 5m KV cache Video ・ 3m Ungraded Lab: KV Cache Code Example ・ 10m Flash attention Video ・ 6m Visualization: Flash Attention Code Example ・ 10m Speculative decoding Video ・ 6m Visualization: Speculative Decoding Code Example ・ 1h Issues that arise in production Video ・ 5m Additional Readings for Module 3 Reading ・ 10m Module 3: Graded Lab Graded ・Code Assignment ・ 1h Module 3: Quiz Graded ・Quiz ・ 30m Acknowledgements Reading ・ 1m · Source
What the credential is
- Credential type
- Course certificate with Pro · www.deeplearning.ai / transformers-in-practice
- Credential
- Course certificate with Pro · www.deeplearning.ai / transformers-in-practice
- Exam requirement
- Three graded labs and three quizzes are listed; no separate final examination is stated. · www.deeplearning.ai / transformers-in-practice
- Credential scope
- As checked on 15 October 2026, the course page identifies a certificate for Pro members who complete the course. The capture does not establish an independent credential issuer, full completion thresholds, expiry, renewal or academic-credit eligibility. · www.deeplearning.ai / transformers-in-practice
Source differences retained
As checked on 15 October 2026, the official page gives differing information on the points below. Each reading is kept as found; none is reconciled.
- Credential: The course description promises a certificate recognizing transformer-based language-model skills. A FAQ answer instead says AI prompting, apparently reused wording. This page records the course-specific certificate, keeps this discrepancy and infers no prompting credential. · Source
- Duration: Visible course duration is 11h1m; same-course embedded metadata has generic courseWorkload PT1H. The visible duration is retained as text; no single numeric completion estimate is used. · Source
- Price: Embedded Course metadata categorizes an offer as Free. The visible page separately requires Pro for a certificate. This does not establish free credential access. No price from an older provider-wide snapshot is inherited. · Source
- Language: As checked on 15 October 2026, the same-course structured metadata states inLanguage="en" and availableLanguage=["en","jp","kr"]. English is retained; the additional literal codes do not establish whether they describe teaching audio, subtitles or another language option. No additional delivery language is verified. · Source
Alternatives in the registry
Related by topic or level.
- Fast & Efficient LLM Inference with vLLM : Check official price; 1h38m listed course duration; individual activity estimates are rounded (duration checked 15 October 2026).
- Intro to LLMs : Check official price; 3 hours (provider headline; module and exam timings listed separately) (duration checked 27 September 2026).
- Finetuning with Llama : Check official price; 2 hours (provider headline; section and exam times listed separately) (duration checked 25 September 2026).
Change log
- : Initial source review from the dated US capture. · Source
See price and enroll at DeepLearning.AI (official site) ↗
Official site. No commission.