# Host-side files

Upstream: [google/embeddinggemma-2](https://huggingface.co/google/embeddinggemma-2)
Revision: `914f7f89142e33e77833254d9c9b90c3cef7303b`
Tree: https://huggingface.co/google/embeddinggemma-2/tree/914f7f89142e33e77833254d9c9b90c3cef7303b

Apache-2.0 allows this redistribution. Files listed as **copied verbatim**
are byte-identical to the upstream objects. The embedding table is the
exception: it is **extracted** from `model.safetensors` tensor
`language_model.embed_tokens.weight` (BF16 `[262144, 512]`) and stored as
`embed_tokens.safetensors` with Gemma's `embed_scale = sqrt(512)` ≈ 22.627417
recorded beside the raw matrix. The full 1.49 GB `model.safetensors` is
**not** mirrored.

See `LICENSE` and `NOTICE` in this folder (and at the repo root).

| File | Origin | Upstream | SHA256 | Bytes |
| --- | --- | --- | --- | --- |
| `config.json` | copied verbatim | `google/embeddinggemma-2@914f7f89142e33e77833254d9c9b90c3cef7303b/config.json` | `b8f1e9931b57fbc054acdb445c41765d55b0074c58d145fa82839941ad1b5bb3` | 4455 |
| `config_sentence_transformers.json` | copied verbatim | `google/embeddinggemma-2@914f7f89142e33e77833254d9c9b90c3cef7303b/config_sentence_transformers.json` | `031e56a498d33c349ab489a21885bcfe25b4fcba841149dc99e1e90d4a7c28f5` | 1565 |
| `tokenizer.json` | copied verbatim | `google/embeddinggemma-2@914f7f89142e33e77833254d9c9b90c3cef7303b/tokenizer.json` | `4d777ef5bdc1aa36227abdfb77c3e49e7b9c892d16e1b6bda41c393504828be4` | 32170510 |
| `tokenizer.model` | copied verbatim | `google/embeddinggemma-2@914f7f89142e33e77833254d9c9b90c3cef7303b/tokenizer.model` | `e594c8a90eb08d8bda498ff4747977dc827ae0c3c56b5c0d41a605a22d02ef03` | 4689013 |
| `tokenizer_config.json` | copied verbatim | `google/embeddinggemma-2@914f7f89142e33e77833254d9c9b90c3cef7303b/tokenizer_config.json` | `17bd5d6e9364ca49a534e1502076593317c298d4a663623091ed45388f004874` | 1599 |
| `preprocessor_config.json` | copied verbatim | `google/embeddinggemma-2@914f7f89142e33e77833254d9c9b90c3cef7303b/preprocessor_config.json` | `ea2ae257e901064abdd98dceb19f2b0da06af600bed15e0f99f5c85c37ee9d78` | 511 |
| `processor_config.json` | copied verbatim | `google/embeddinggemma-2@914f7f89142e33e77833254d9c9b90c3cef7303b/processor_config.json` | `168f6a08522f3ce5dea596d94d003af2fd691742d4f41fe1f9d8cce76bfbf69c` | 1788 |
| `chat_template.jinja` | copied verbatim | `google/embeddinggemma-2@914f7f89142e33e77833254d9c9b90c3cef7303b/chat_template.jinja` | `4b852efc0b9960283e735363331e6f325b33bc74bdbaa076f595bc4e9b94d85e` | 1016 |
| `embed_tokens.safetensors` | extracted | `google/embeddinggemma-2@914f7f89142e33e77833254d9c9b90c3cef7303b/model.safetensors (language_model.embed_tokens.weight)` | `a37078ff5786d9bf75b059bdc8f4009f4311bf8e7ffbc287b66bab3d38457996` | 268435637 |
| `LICENSE` | added for redistribution | `anemll-embeddings hf/LICENSE (Apache-2.0 / attribution)` | `a6cba85bc92e0cff7a450b1d873c0eaa2e9fc96bf472df0247a26bec77bf3ff9` | 10173 |
| `NOTICE` | added for redistribution | `anemll-embeddings hf/NOTICE (Apache-2.0 / attribution)` | `afc64bd52b69b95ce84dd35cfd7a382fe0be746238be4243d3ae463f567962a3` | 1821 |
| `SHA256SUMS` | added for redistribution | `generated by scripts/prepare_hf_host_folder.py` | `3a41a4f05dffe3ec1a67427310ab739ea7f73e507ab566d4da8ec303a349dfe1` | 932 |