| --- |
| library_name: mlx |
| license: eupl-1.2 |
| pipeline_tag: image-text-to-text |
| base_model_relation: quantized |
| tags: |
| - gemma4 |
| - mlx |
| - apple-silicon |
| - 4bit |
| - on-device |
| - conversational |
| base_model: |
| - google/gemma-4-E2B-it |
| --- |
| |
| # LetheanNetwork/lemer-mlx |
|
|
| Gemma 4 E2B in MLX format, 4-bit quantized, converted from |
| [LetheanNetwork/lemer](https://huggingface.co/LetheanNetwork/lemer)'s |
| bf16 safetensors via `mlx_lm.convert`. This is the unmodified Google |
| Gemma 4 E2B-IT weights — no LEK shift, no fine-tuning — hosted in our |
| namespace so downstream tools (benchmarks, apps) don't have to depend |
| on external mlx-community mirrors. |
|
|
| For the LEK-merged (consent-based ethical kernel) variant of the same |
| model, see [`lthn/lemer`](https://huggingface.co/lthn/lemer). |
|
|
| ## Variants in this family |
|
|
| | Repo | Format | Bits | Use case | |
| |---|---|---|---| |
| | [`LetheanNetwork/lemer`](https://huggingface.co/LetheanNetwork/lemer) | safetensors + gguf Q4_K_M | bf16 / 4 | Source weights + llama.cpp/Ollama | |
| | **`LetheanNetwork/lemer-mlx`** | mlx | 4 | **This repo** — Apple Silicon default | |
| | [`LetheanNetwork/lemer-mlx-8bit`](https://huggingface.co/LetheanNetwork/lemer-mlx-8bit) | mlx | 8 | Apple Silicon higher-precision | |
| | [`LetheanNetwork/lemer-mlx-bf16`](https://huggingface.co/LetheanNetwork/lemer-mlx-bf16) | mlx | bf16 | Apple Silicon full-precision reference | |
|
|
| ## Usage |
|
|
| ```python |
| from mlx_lm import load, generate |
| |
| model, tokenizer = load("LetheanNetwork/lemer-mlx") |
| response = generate( |
| model, tokenizer, |
| prompt=tokenizer.apply_chat_template( |
| [{"role": "user", "content": "Hello"}], |
| add_generation_prompt=True, |
| enable_thinking=True, |
| ), |
| max_tokens=512, |
| ) |
| ``` |
|
|
| ## Provenance |
|
|
| - Source: `LetheanNetwork/lemer` bf16 safetensors (= `google/gemma-4-E2B-it`) |
| - Converter: `mlx_lm.convert` (mlx-lm — LM Studio / Apple ML Research) |
| - Quant: 4-bit group quantization, ~4.5 bits/weight effective |
| - License: Apache 2.0 (Gemma Terms of Use) |
|
|
| ## License |
|
|
| Apache 2.0, subject to the [Gemma Terms of Use](https://ai.google.dev/gemma/docs/gemma_4_license). |
