Ben Graville commited on
Upload README.md with huggingface_hub
Browse files
README.md
ADDED
|
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: coreai
|
| 3 |
+
tags: [core-ai, aimodel, on-device, apple]
|
| 4 |
+
---
|
| 5 |
+
|
| 6 |
+
# LFM2.5-2.6B-CoreAI
|
| 7 |
+
|
| 8 |
+
Core AI `.aimodel` bundles converted for [Visible](https://visible.cx) on-device
|
| 9 |
+
inference. **Derivative artifacts** — converted from the upstream checkpoints listed
|
| 10 |
+
below; all original model licences apply and travel with these weights.
|
| 11 |
+
|
| 12 |
+
| Bundle | Source checkpoint | Context | Functions | Producer | Created |
|
| 13 |
+
|---|---|---|---|---|---|
|
| 14 |
+
| `gpu-pipelined/lfm2_5_2_6b_decode_int8hu_block32_sym_mf64` | LiquidAI/LFM2.5-2.6B | 4096 | main+prefill | coreai-core 1.0.0b2 | 20260817T111106Z |
|
| 15 |
+
|
| 16 |
+
## Provenance
|
| 17 |
+
|
| 18 |
+
- Converted with the [coreai-model-zoo](https://github.com/john-rocky/coreai-model-zoo)
|
| 19 |
+
conversion recipes on the pinned `apple/coreai-models` base
|
| 20 |
+
`b1cb71b8522d99408059fa0b98b8742171bcb0b8` + the zoo python overlay.
|
| 21 |
+
- Toolchain: `coreai-torch 0.4.1`, `coreai-core 1.0.0b2`, `coreai-opt 0.2.1`, `torch 2.9.0`.
|
| 22 |
+
- Every bundle's inner `<name>.aimodel/metadata.json` reports
|
| 23 |
+
`producer: "coreai-core 1.0.0b2"`.
|
| 24 |
+
- Gemma bundles: `eos_token` is set to `<turn|>` (Gemma 4 ends a *turn* with `<turn|>`;
|
| 25 |
+
stopping on the source's `<eos>` overruns the reply).
|
| 26 |
+
- Bundles carry `runtime_env COREAI_CHUNK_THRESHOLD=1`.
|
| 27 |
+
|
| 28 |
+
Upstream sources: `LiquidAI/LFM2.5-2.6B`
|