eka-ai 1.0.0__tar.gz

eka_ai-1.0.0/LICENSE

@@ -0,0 +1,159 @@
+Apache License
+Version 2.0, January 2004
+http://www.apache.org/licenses/
+
+TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+1. Definitions.
+
+   "License" shall mean the terms and conditions for use, reproduction,
+   and distribution as defined by Sections 1 through 9 of this document.
+
+   "Licensor" shall mean the copyright owner or entity authorized by
+   the copyright owner that is granting the License.
+
+   "Legal Entity" shall mean the union of the acting entity and all
+   other entities that control, are controlled by, or are under common
+   control with that entity. For the purposes of this definition,
+   "control" means (i) the power, direct or indirect, to cause the
+   direction or management of such entity, whether by contract or
+   otherwise, or (ii) ownership of fifty percent (50%) or more of the
+   outstanding shares, or (iii) beneficial ownership of such entity.
+
+   "You" (or "Your") shall mean an individual or Legal Entity
+   exercising permissions granted by this License.
+
+   "Source" form shall mean the preferred form for making modifications,
+   including but not limited to software source code, documentation
+   source, and configuration files.
+
+   "Object" form shall mean any form resulting from mechanical
+   transformation or translation of a Source form, including but
+   not limited to compiled object code, generated documentation,
+   and conversions to other media types.
+
+   "Work" shall mean the work of authorship made available under
+   the License, as indicated by a copyright notice that is included in
+   or attached to the work (an example is provided in the Appendix below).
+
+   "Derivative Works" shall mean any work, whether in Source or Object
+   form, that is based on (or derived from) the Work and for which the
+   editorial revisions, annotations, elaborations, or other modifications
+   represent, as a whole, an original work of authorship. For the purposes
+   of this License, Derivative Works shall not include works that remain
+   separable from, or merely link (or bind by name) to the interfaces of,
+   the Work and Derivative Works thereof.
+
+   "Contribution" shall mean, as submitted to the Licensor for inclusion
+   in the Work by the copyright owner or by an individual or Legal Entity
+   authorized to submit on behalf of the copyright owner.
+
+   "Contributor" shall mean Licensor and any Legal Entity on behalf of
+   whom a Contribution has been received by the Licensor and included
+   within the Work.
+
+2. Grant of Copyright License. Subject to the terms and conditions of
+   this License, each Contributor hereby grants to You a perpetual,
+   worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+   copyright license to reproduce, prepare Derivative Works of,
+   publicly display, publicly perform, sublicense, and distribute the
+   Work and such Derivative Works in Source or Object form.
+
+3. Grant of Patent License. Subject to the terms and conditions of
+   this License, each Contributor hereby grants to You a perpetual,
+   worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+   (except as stated in this section) patent license to make, have made,
+   use, offer to sell, sell, import, and otherwise transfer the Work,
+   where such license applies only to those patent claims licensable
+   by such Contributor that are necessarily infringed by their
+   Contribution(s) alone or by the combination of their Contribution(s)
+   with the Work to which such Contribution(s) was submitted.
+
+4. Redistribution. You may reproduce and distribute copies of the
+   Work or Derivative Works thereof in any medium, with or without
+   modifications, and in Source or Object form, provided that You
+   meet the following conditions:
+
+   (a) You must give any other recipients of the Work or
+       Derivative Works a copy of this License; and
+
+   (b) You must cause any modified files to carry prominent notices
+       stating that You changed the files; and
+
+   (c) You must retain, in the Source form of any Derivative Works
+       that You distribute, all copyright, patent, trademark, and
+       attribution notices from the Source form of the Work,
+       excluding those notices that do not pertain to any part of
+       the Derivative Works; and
+
+   (d) If the Work includes a "NOTICE" text file as part of its
+       distribution, You must include a readable copy of the
+       attribution notices contained within such NOTICE file, in
+       at least one of the following places: within a NOTICE text
+       file distributed as part of the Derivative Works; within
+       the Source form or documentation, if provided along with the
+       Derivative Works; or, within a display generated by the
+       Derivative Works, if and wherever such third-party notices
+       normally appear.
+
+   The contents of the NOTICE file are for informational purposes
+   only and do not modify the License.
+
+5. Submission of Contributions. Unless You explicitly state otherwise,
+   any Contribution intentionally submitted for inclusion in the Work
+   by You to the Licensor shall be under the terms and conditions of
+   this License, without any additional terms or conditions.
+
+6. Trademarks. This License does not grant permission to use the trade
+   names, trademarks, service marks, or product names of the Licensor,
+   except as required for reasonable and customary use in describing the
+   origin of the Work.
+
+7. Disclaimer of Warranty. Unless required by applicable law or
+   agreed to in writing, Licensor provides the Work (and each
+   Contributor provides its Contributions) on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+   implied, including, without limitation, any warranties or conditions
+   of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+   PARTICULAR PURPOSE. You are solely responsible for determining the
+   appropriateness of using or reproducing the Work and assume any
+   risks associated with Your exercise of permissions under this License.
+
+8. Limitation of Liability. In no event and under no legal theory,
+   whether in tort (including negligence), contract, or otherwise,
+   unless required by applicable law (such as deliberate and grossly
+   negligent acts) or agreed to in writing, shall any Contributor be
+   liable to You for damages, including any direct, indirect, special,
+   incidental, or exemplary damages of any character arising as a
+   result of this License or out of the use or inability to use the
+   Work (including but not limited to damages for loss of goodwill,
+   work stoppage, computer failure or malfunction, or all other
+   commercial damages or losses), even if such Contributor has been
+   advised of the possibility of such damages.
+
+9. Accepting Warranty or Additional Liability. While redistributing
+   the Work or Derivative Works thereof, You may choose to offer,
+   and charge a fee for, acceptance of support, warranty, indemnity,
+   or other liability obligations and/or rights consistent with this
+   License. However, in accepting such obligations, You may act only
+   on Your own behalf and on Your sole responsibility, not on behalf
+   of any other Contributor, and only if You agree to indemnify,
+   defend, and hold each Contributor harmless for any liability
+   incurred by, or claims asserted against, such Contributor by reason
+   of your accepting any such warranty or additional liability.
+
+END OF TERMS AND CONDITIONS
+
+Copyright 2024 EKA-AI Contributors
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.

eka_ai-1.0.0/MANIFEST.in

@@ -0,0 +1,11 @@
+include README.md
+include LICENSE
+include requirements.txt
+include pyproject.toml
+recursive-include eka_ai *.py *.typed
+recursive-include tests *.py
+recursive-include examples *.py
+global-exclude __pycache__
+global-exclude *.py[cod]
+global-exclude *.so
+global-exclude .DS_Store

eka_ai-1.0.0/PKG-INFO

@@ -0,0 +1,424 @@
+Metadata-Version: 2.4
+Name: eka-ai
+Version: 1.0.0
+Summary: A 109M parameter historical language model trained from scratch on Project Gutenberg
+Author: EKA-AI Contributors
+License: Apache-2.0
+Project-URL: Homepage, https://github.com/eka-ai/eka-ai
+Project-URL: Repository, https://github.com/eka-ai/eka-ai
+Project-URL: Issues, https://github.com/eka-ai/eka-ai/issues
+Project-URL: Changelog, https://github.com/eka-ai/eka-ai/blob/main/CHANGELOG.md
+Keywords: language-model,historical,transformer,nlp,pytorch,text-generation,gutenberg
+Classifier: Development Status :: 5 - Production/Stable
+Classifier: Intended Audience :: Developers
+Classifier: Intended Audience :: Science/Research
+Classifier: License :: OSI Approved :: Apache Software License
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.9
+Classifier: Programming Language :: Python :: 3.10
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
+Classifier: Topic :: Text Processing :: Linguistic
+Requires-Python: >=3.9
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: torch>=2.1.0
+Requires-Dist: sentencepiece>=0.1.99
+Requires-Dist: gdown>=4.7.3
+Provides-Extra: dev
+Requires-Dist: pytest>=7.4; extra == "dev"
+Requires-Dist: pytest-cov>=4.1; extra == "dev"
+Requires-Dist: black>=23.0; extra == "dev"
+Requires-Dist: isort>=5.12; extra == "dev"
+Requires-Dist: ruff>=0.1.0; extra == "dev"
+Requires-Dist: mypy>=1.5; extra == "dev"
+Provides-Extra: docs
+Requires-Dist: sphinx>=7.0; extra == "docs"
+Requires-Dist: sphinx-rtd-theme>=1.3; extra == "docs"
+Requires-Dist: myst-parser>=2.0; extra == "docs"
+Dynamic: license-file
+
+<div align="center">
+
+# EKA-1
+
+**A 109M parameter historical language model trained from scratch on Project Gutenberg**
+
+[![PyPI version](https://img.shields.io/pypi/v/eka-ai.svg)](https://pypi.org/project/eka-ai/)
+[![Python](https://img.shields.io/pypi/pyversions/eka-ai.svg)](https://pypi.org/project/eka-ai/)
+[![License](https://img.shields.io/badge/license-Apache%202.0-blue.svg)](LICENSE)
+[![Tests](https://github.com/eka-ai/eka-ai/actions/workflows/tests.yml/badge.svg)](https://github.com/eka-ai/eka-ai/actions/workflows/tests.yml)
+
+</div>
+
+---
+
+EKA-1 is a **decoder-only Transformer** with **109,529,856 parameters**, trained entirely from scratch on classical texts from [Project Gutenberg](https://www.gutenberg.org/). It features a modern architecture — RMSNorm, Rotary Position Embeddings (RoPE), SwiGLU activation, multi-head causal attention via PyTorch SDPA, and weight tying — packed into a library with a clean, one-line API.
+
+---
+
+## Table of Contents
+
+- [Features](#features)
+- [Installation](#installation)
+- [Quick Start](#quick-start)
+- [API Reference](#api-reference)
+- [Model Architecture](#model-architecture)
+- [Model Configuration](#model-configuration)
+- [Examples](#examples)
+- [Development](#development)
+- [Publishing to PyPI](#publishing-to-pypi)
+- [License](#license)
+
+---
+
+## Features
+
+| Feature | Status |
+|---|---|
+| Automatic model & tokenizer download | ✅ |
+| CPU inference | ✅ |
+| CUDA inference | ✅ |
+| Text generation | ✅ |
+| Chat interface (single & multi-turn) | ✅ |
+| Streaming generation | ✅ |
+| Temperature sampling | ✅ |
+| Top-k sampling | ✅ |
+| Top-p (nucleus) sampling | ✅ |
+| Repetition penalty | ✅ |
+| Context truncation | ✅ |
+| PEP 561 typed package | ✅ |
+
+---
+
+## Installation
+
+**From PyPI (recommended):**
+
+```bash
+pip install eka-ai
+```
+
+**From source:**
+
+```bash
+git clone https://github.com/eka-ai/eka-ai.git
+cd eka-ai
+pip install -e ".[dev]"
+```
+
+**Requirements:**
+
+- Python ≥ 3.9
+- PyTorch ≥ 2.1
+- sentencepiece ≥ 0.1.99
+- gdown ≥ 4.7.3
+
+On first use, `EKA()` automatically downloads `eka_model.pt` (~424 MB) and `tokenizer.model` (~768 KB) to `~/.cache/eka_ai/`.
+
+> **Custom cache location:** Set the `EKA_CACHE_DIR` environment variable to override the default cache directory.
+
+---
+
+## Quick Start
+
+```python
+from eka_ai import EKA
+
+# Loads model on first call — downloads files automatically
+model = EKA()
+
+# ── Text generation ────────────────────────────────────────
+print(
+    model.generate(
+        "Tell me about the Roman Empire",
+        max_new_tokens=128,
+    )
+)
+
+# ── Single-turn chat ───────────────────────────────────────
+print(
+    model.chat(
+        "Who was Ashoka?"
+    )
+)
+
+# ── Streaming generation ───────────────────────────────────
+for token in model.stream(
+    "Tell me a story"
+):
+    print(token, end="", flush=True)
+```
+
+---
+
+## API Reference
+
+### `EKA(...)` — constructor
+
+```python
+model = EKA(
+    device=None,           # "cpu" | "cuda" | torch.device — auto-detected
+    dtype=None,            # torch.bfloat16 on CPU by default
+    model_path=None,       # override cached model path
+    tokenizer_path=None,   # override cached tokenizer path
+    auto_download=True,    # download files if not cached
+    system_prompt=None,    # custom system prompt for chat()
+    verbose=True,          # print loading progress
+)
+```
+
+---
+
+### `model.generate(prompt, ...)` → `str`
+
+Generate a text continuation for a plain prompt.
+
+```python
+text = model.generate(
+    prompt,
+    max_new_tokens=256,        # int   — max tokens to generate
+    temperature=0.8,           # float — sampling temperature (> 0)
+    top_k=50,                  # int   — top-k cutoff (0 = disabled)
+    top_p=0.95,                # float — nucleus threshold (1.0 = disabled)
+    repetition_penalty=1.1,    # float — penalty for repeated tokens (1.0 = disabled)
+    return_result=False,       # bool  — return GenerationResult with metadata
+)
+```
+
+When `return_result=True`, returns a `GenerationResult` with attributes:
+- `.text` — the generated text
+- `.tokens_generated` — number of new tokens
+- `.tokens_per_second` — generation throughput
+- `.device` — device used
+
+---
+
+### `model.chat(message, ...)` → `str`
+
+Single-turn or multi-turn chat using the built-in chat template.
+
+```python
+reply = model.chat(
+    message,
+    history=None,              # list of (user, assistant) tuples
+    system_prompt=None,        # override instance system prompt
+    max_new_tokens=256,
+    temperature=0.8,
+    top_k=50,
+    top_p=0.95,
+    repetition_penalty=1.1,
+)
+```
+
+**Multi-turn example:**
+
+```python
+history = []
+reply1 = model.chat("Who was Julius Caesar?")
+history.append(("Who was Julius Caesar?", reply1))
+
+reply2 = model.chat("What were his greatest conquests?", history=history)
+```
+
+---
+
+### `model.stream(prompt, ...)` → `Iterator[str]`
+
+Streaming generation — yields one decoded token piece at a time.
+
+```python
+for piece in model.stream(
+    "The ancient library of",
+    max_new_tokens=128,
+    temperature=0.8,
+):
+    print(piece, end="", flush=True)
+```
+
+---
+
+### `model.stream_chat(message, ...)` → `Iterator[str]`
+
+Like `stream()` but uses the chat template.
+
+```python
+for piece in model.stream_chat("Tell me about Socrates"):
+    print(piece, end="", flush=True)
+```
+
+---
+
+### `model.info()` → `dict`
+
+```python
+info = model.info()
+# {
+#   "version": "1.0.0",
+#   "parameters": 109529856,
+#   "device": "cpu",
+#   "dtype": "torch.bfloat16",
+#   "vocab_size": 32000,
+#   "n_layers": 12,
+#   "n_heads": 12,
+#   "d_model": 768,
+#   "context_length": 512,
+# }
+```
+
+---
+
+## Model Architecture
+
+EKA-1 is a decoder-only Transformer with the following design choices:
+
+| Component | Implementation |
+|---|---|
+| Normalisation | **RMSNorm** (no bias, no mean subtraction) |
+| Position encoding | **Rotary Position Embeddings (RoPE)** — no learned position table |
+| Feed-forward | **SwiGLU** — `down(silu(gate(x)) ⊙ up(x))` |
+| Attention | **Multi-Head Causal Attention** via PyTorch `scaled_dot_product_attention` |
+| Weight sharing | **Weight tying** — embedding and LM-head share parameters |
+| Architecture | Pre-norm residual blocks |
+| Bias | None (all linear layers are bias-free) |
+
+The attention implementation uses PyTorch's built-in SDPA which automatically selects the optimal kernel: FlashAttention 2 (when available), memory-efficient attention, or the reference implementation.
+
+---
+
+## Model Configuration
+
+```json
+{
+  "vocab_size": 32000,
+  "n_layers": 12,
+  "n_heads": 12,
+  "n_kv_heads": 12,
+  "d_model": 768,
+  "d_ffn": 3072,
+  "context_length": 512,
+  "dropout": 0.0,
+  "bias": false
+}
+```
+
+| Parameter | Value |
+|---|---|
+| Total parameters | **109,529,856** |
+| Embedding dimension | 768 |
+| Attention heads | 12 × 64-dim heads |
+| Transformer layers | 12 |
+| FFN hidden dim | 2048 (after SwiGLU 2/3 ratio) |
+| Vocabulary | 32,000 BPE tokens (SentencePiece) |
+| Context length | 512 tokens |
+| Training data | Project Gutenberg |
+
+---
+
+## Examples
+
+Run the provided example scripts from the repository root:
+
+```bash
+# Interactive text generation
+python examples/generation.py --prompt "The fall of Rome" --max_tokens 200
+
+# Interactive chat REPL
+python examples/chat.py
+python examples/chat.py --stream       # streaming mode
+python examples/chat.py --device cuda  # GPU
+
+# Throughput benchmark
+python examples/benchmark.py --runs 10 --max_tokens 256
+python examples/benchmark.py --device cuda
+```
+
+---
+
+## Development
+
+```bash
+# Clone and install in editable mode with dev extras
+git clone https://github.com/eka-ai/eka-ai.git
+cd eka-ai
+pip install -e ".[dev]"
+
+# Run the full test suite
+pytest tests/ -v
+
+# Run specific test file
+pytest tests/test_model.py -v
+
+# Lint
+ruff check eka_ai/ tests/
+black --check eka_ai/ tests/ examples/
+
+# Auto-format
+black eka_ai/ tests/ examples/
+isort eka_ai/ tests/
+
+# Type check
+mypy eka_ai/
+```
+
+### Project Structure
+
+```
+eka-ai/
+│
+├── eka_ai/
+│   ├── __init__.py       # Public API exports
+│   ├── config.py         # EKAConfig dataclass
+│   ├── model.py          # EKA1Model architecture
+│   ├── tokenizer.py      # EKATokenizer (SentencePiece wrapper)
+│   ├── downloader.py     # Automatic model/tokenizer download
+│   ├── generation.py     # EKA high-level API
+│   ├── utils.py          # Sampling utilities
+│   └── py.typed          # PEP 561 marker
+│
+├── examples/
+│   ├── chat.py           # Interactive chat REPL
+│   ├── generation.py     # Text generation demo
+│   └── benchmark.py      # Throughput benchmark
+│
+├── tests/
+│   ├── test_config.py    # Config unit tests
+│   ├── test_model.py     # Architecture unit tests
+│   ├── test_utils.py     # Sampling utility tests
+│   └── test_generation.py# API integration tests
+│
+├── .github/workflows/
+│   ├── publish.yml       # PyPI publishing workflow
+│   └── tests.yml         # CI test matrix
+│
+├── README.md
+├── LICENSE               # Apache 2.0
+├── setup.py
+├── pyproject.toml
+├── requirements.txt
+└── MANIFEST.in
+```
+
+
+## Citation
+
+If you use EKA-1 in your research, please cite:
+
+```bibtex
+@misc{eka1_2024,
+  title  = {{EKA-1}: A 109M Parameter Historical Language Model},
+  author = {chvkrsubhash},
+  year   = {2026},
+  url    = {https://github.com/eka-ai/eka-ai},
+  note   = {Trained from scratch on Project Gutenberg}
+}
+```
+
+---
+
+## License
+
+This project is licensed under the **Apache License 2.0** — see [LICENSE](LICENSE) for details.
+
+The training data is sourced from [Project Gutenberg](https://www.gutenberg.org/), which consists of public-domain works.