- Renamed Decoder (and decoder.py) to GptDecoder (gpt_decoder.py) for clarity in GPT1
- Implemented support for cache and use_cache parameters in GptDecoder.forward (API unification)
- Adapted all usages in GPT model to use new decoder structure and handle tuple output
- Refactored core tests (test_gpt.py, test_gpt_decoder.py, test_basic.py) to correctly expect tuple or logits and ensure shape/device checks work as before
- Improved clarity and future extensibility for autoregressive generation and benchmarking
- No changes to architectural details or training loop; pure API and test modernization
- Add new core modules: GeGLU (Gated GELU Linear Unit), GemmaDecoder, MultiQueryAttention; all with highly detailed scientific (RU) docstrings: theory, usage, formulas, references
- Major doc improvements in Gemma model: class, __init__, forward, generate now have full educational/engineering docstrings, use-case samples, and literature links
- Add comprehensive unit tests:
* tests/core/test_geglu.py: GeGLU coverage (shape, grads, edge, repeat, float16/skip)
* tests/core/test_gemma_decoder.py: GemmaDecoder coverage (shape, mask, cache, repeatability, errors)
* tests/core/test_multi_query_attention.py: MQA coverage (shape, cache, gradients, masking, dropout, raise)
- All modules and tests follow strict quality/documentation standards, code is now robust for research & production
- Add core Gemma model (architecture, attention, GeGLU, RoPE, RMSNorm, etc)
- Add configs for training and generation: gemma_train.json, gemma_generate.json
- Add Gemma notebook for exploratory analysis and demonstration
- Add __init__.py for Gemma submodule
- Update run_llm_experiment.py to support Gemma experiment configs
test(gemma): add comprehensive unit tests for Gemma
- Test forward pass (with/without cache)
- Test autoregressive generation (greedy, top-k, top-p)
- Test shape correctness and max sequence length errors
- Test multi-layer stack and token embeddings
docs: add documentation notebook for Gemma usage and analysis
Closes: #issue (if applicable)
- Implement new core module: MixtralDecoder (llm/core/mixtral_decoder.py) with full Russian scientific docstrings, formal math, and usage examples
- Improve MoE: add Russian docstrings for class, __init__, forward; validate top_k_experts; explain theory and components
- Refactor Mixtral model: switch stack to MixtralDecoder, add comprehensive documentation for class, constructor and forward, clarify config usage and architecture
- Add thorough unit tests:
* tests/core/test_mixtral_decoder.py: checks shapes, errors, mask, dropout, grads etc.
* tests/core/test_moe.py: covers normal and edge-case logic, gradients, shape, params check
- All code and tests in compliance with recent scientific and engineering standards.
- Add Mixtral architecture implementation with MoE support (llm/src/llm/models/mixtral/mixtral.py)
- Introduce generic Mixture-of-Experts (MoE) block (llm/src/llm/core/moe.py)
- Create dedicated configuration files for Mixtral training and generation experiments
- Register and test Mixtral support in experiment runner (run_llm_experiment.py)
- Add unit tests for Mixtral API including forward, caching, and generation modes
- Include Jupyter notebook mixstral.ipynb for architectural exploration and research
- Ensure correct handling of torch bool masks in sampling (top-k, top-p) during generation
BREAKING CHANGE: Adds new model code and test coverage, modifying experiment runner logic to register Mixtral.
- Expanded module-level and function/class docstrings in optimizer.py, scheduler.py, and trainer.py
- Described mathematical foundations, theoretical motivations, and provided detailed usage examples for students
- All docstrings in Russian, clear scientific style
test(training): add comprehensive tests for optimizer, scheduler, and trainer modules
- Added new test files for get_optimizer, get_linear_schedule_with_warmup, and Trainer
- Tests cover parameter handling, edge cases, and expected learning dynamics (lr schedules and loss behavior)
- Trainer now logs average epoch losses to self.loss_history for testability and analysis
refactor(training/trainer): log epoch loss to loss_history for downstream analysis and tests
BREAKING CHANGE: Trainer.loss_history is a new attribute consolidating average losses per epoch, enabling robust learning dynamics assertions in tests
- docs: rewrite and expand docstrings for SwiGLU class and forward method (motivation, math, architecture, usage, references to LLaMA/Mistral/PaLM)
- test: add unit tests for SwiGLU (shape, dtype, gradients, output range, fp16 support, reproducibility)
- strictly doc/tests, no logic or API changes
This improves transparency and reliability for gated FFN blocks in transformer architectures.
- docs: expand and clarify docstrings for SiLU class and its method (mathematical formula, motivation, properties vs ReLU/GELU, usage, and references to Swish/LLM papers)
- test: add unit tests for SiLU (shape/dtype, behavior on large/small values, PyTorch reference, gradients, broadcast)
- no logic/API changes
This update improves reliability and usability of the SiLU activation module.
- docs: update/increase docstring detail for RMSNorm class and methods (motivation, formula, architecture, usage, references to LLaMA/PaLM/GPT)
- test: add comprehensive unit tests for RMSNorm (shape/type preservation, rms scaling, gradients for input and weights, fp16, large eps stability)
No code/API changes beyond docs and new tests.
- docs: rewrite and expand docstrings for GELU class and method (motivation, math formula, smoother ReLU for Transformers, usage, references)
- test: add dedicated tests for GELU (output shape, dtype, comparison with torch GELU, monotonicity, gradients, large/small value behavior)
- fix: align numerical test to allow for minor approximation difference vs PyTorch gelu
This update makes the GELU module more transparent and robust for deep learning practitioners and researchers.
- docs: Add detailed docstrings for CachedDecoder class and its methods (__init__, forward); explain autoregressive caching, architecture, math, usage, and links to GPT-2/LLM references
- test: Add comprehensive unit tests for CachedDecoder (initialization, forward with and without cache, cache chaining, output shape, error on long input, backward pass)
- These changes improve code clarity, reliability, and testing for decoder blocks with KV cache.
- docs: expanded docstrings for MistralDecoder class and methods (__init__, forward); explained architecture, key parameters, usage, and links to relevant papers (Mistral, Llama 2)
- test: add comprehensive unit tests for MistralDecoder (init, forward, cache handling, output shape, shape errors, backward)
- These changes improve explainability, reliability, and test coverage for the decoder module.
- docs: Rewrite and expand docstrings for the GroupedQueryAttention class and all main methods (__init__, forward, _repeat_kv_heads, _create_sliding_window_mask):
- explained GQA architecture and motivation
- included mathematical formulas, step-by-step algorithms, usage examples
- added references to relevant scientific papers (Mistral, Llama 2, etc.)
- test: Add dedicated unit tests for GQA (output shape correctness, mask/window logic, KV head replication, RoPE processing, error and edge-cases)
- docs/test: Documentation and tests now fully reflect modern GQA usage and best practices for LLM architectures
This commit makes the implementation, usage, and theoretical underpinnings of GQA transparent and reproducible for researchers and engineers.
- Covers inference with and without cache and with sampling (top-k, top-p)
- Includes test for max sequence length (should raise ValueError)
- Verifies output shape and absence of dtype errors for the mask logic
- Minimal config and random data ensure tests are fast and robust
Motivation: Regression and integration protection for Llama decoding and sampling logic.
- Covers forward pass with and without KV-cache
- Verifies correct sequence generation for greedy, top-k, and top-p sampling
- Adds ValueError test for exceeding max sequence length
- Uses small random toy config and minimal setup for fast test feedback
Motivation: Prevent regressions in decoding, sampling, and KV-cache logic in GPT2 implementation.
- refactor: улучшена и унифицирована реализация RoPE, теперь поддерживаются строгие проверки размерности входа; внесены улучшения и структурные изменения в MultiHeadAttention (более понятная логика, строгая спецификация входов/выходов)
- docs: полностью переписаны docstrings для RoPE и MultiHeadAttention — включены математические формулы, ссылки на научные статьи, подробные пояснения по алгоритму, формату входных данных, ограничениям, примеры использования
- test: добавлены отдельные unit-тесты для RoPE (корректность формы, ошибки на неверную размерность, сохранение нормы, backward/градиенты, работу с параметрами start_pos и батчами)
- chore: удалён неиспользуемый модуль core/head_attention.py
- fix: теперь выбрасывается AssertionError при неправильной размерности входа RoPE; это позволило полностью покрыть тест-кейсы на ошибки
Этот коммит синхронизирует логику реализации базового внимания с современной практикой LLM, укрепляет документацию для инженеров и исследователей, а также расширяет надежность автотестирования библиотеки.
