mlboydaisuke commited on
Commit
fd7bb96
·
verified ·
1 Parent(s): e854549

Card: the root export is gone from main (revision 216f2841 keeps it)

Browse files
Files changed (1) hide show
  1. README.md +7 -6
README.md CHANGED
@@ -23,9 +23,10 @@ This model has no row on [DeviceMark](https://devicemark.github.io/), the on-dev
23
  <!-- gen-cards:devicemark end -->
24
 
25
  > **2026-10-10 re-export, two folders.** `macos-fp16/` and `macos-int8/` were exported with apple/coreai-models
26
- > 97a14be and load with its diffusion pipeline. The files at the repo root are the 2026-07-20 export: current
27
- > apple/coreai-models stops on them with "unsupported metadata_version". No app reads them since 2026-10-10: the zoo's
28
- > CoreAIImageGen app moved to the two folders. They stay for the record.
 
29
 
30
  # FLUX.2 klein 4B — Core AI
31
 
@@ -45,7 +46,7 @@ iPhone 17 Pro, over the ~6.1 GB a 12 GB iPhone gives one app.
45
  | --- | --- | --- | --- | --- |
46
  | `macos-fp16/` | fp16, one asset per graph | text-to-image | 19 | 14,159,882,995 |
47
  | `macos-int8/` | int8 per-block 32 on the text encoder and the transformer, fp16 VAEs | text-to-image, image-to-image | 25 | 7,777,120,667 |
48
- | repo root (2026-07-20) | int4 per-block 32, plus the in-context edit transformers | no app since 2026-10-10 (kept for the record) | 34 | 10,854,216,650 |
49
 
50
  Text-to-image reads 16 files of a folder: 14,090,986,305 bytes from `macos-fp16/`, 7,540,032,209 bytes from
51
  `macos-int8/`. The VAE encoders and the half-size VAEs are only for image-to-image and tiled decoding.
@@ -154,9 +155,9 @@ The text encoder is the Qwen3 model inside the FLUX.2 klein repo: hidden size 25
154
  Qwen3-4B. Earlier versions of this card said 8B; that was wrong. The pipeline reads hidden states 9, 18 and 27, so the
155
  export drops the last 9 layers.
156
 
157
- ## In-context editing (repo root, 2026-07-20)
158
 
159
- The repo root also has **`Transformer_edit.aimodel`**, **`Transformer_edit_512.aimodel`** and
160
  **`Transformer_edit_2ref.aimodel`** for FLUX.2's in-context editing. You give a reference image and an instruction —
161
  *"add a red wizard hat, keep everything else the same"* — and only the instructed change is applied while the
162
  subject, pose, and background are preserved. This is different from strength-based image-to-image (SDEdit), which
 
23
  <!-- gen-cards:devicemark end -->
24
 
25
  > **2026-10-10 re-export, two folders.** `macos-fp16/` and `macos-int8/` were exported with apple/coreai-models
26
+ > 97a14be and load with its diffusion pipeline. The 2026-07-20 export that sat at the repo root (int4, plus the
27
+ > in-context edit transformers) was removed from `main` on 2026-10-10: current apple/coreai-models stops on it with
28
+ > "unsupported metadata_version", and no app reads it since the zoo's CoreAIImageGen app moved to the two folders.
29
+ > It is still there at revision `216f2841`.
30
 
31
  # FLUX.2 klein 4B — Core AI
32
 
 
46
  | --- | --- | --- | --- | --- |
47
  | `macos-fp16/` | fp16, one asset per graph | text-to-image | 19 | 14,159,882,995 |
48
  | `macos-int8/` | int8 per-block 32 on the text encoder and the transformer, fp16 VAEs | text-to-image, image-to-image | 25 | 7,777,120,667 |
49
+ | revision `216f2841` and earlier (2026-07-20; removed from `main` 2026-10-10) | int4 per-block 32, plus the in-context edit transformers | no app | 34 | 10,854,216,650 |
50
 
51
  Text-to-image reads 16 files of a folder: 14,090,986,305 bytes from `macos-fp16/`, 7,540,032,209 bytes from
52
  `macos-int8/`. The VAE encoders and the half-size VAEs are only for image-to-image and tiled decoding.
 
155
  Qwen3-4B. Earlier versions of this card said 8B; that was wrong. The pipeline reads hidden states 9, 18 and 27, so the
156
  export drops the last 9 layers.
157
 
158
+ ## In-context editing (revision `216f2841`, 2026-07-20)
159
 
160
+ Revision `216f2841` and earlier have **`Transformer_edit.aimodel`**, **`Transformer_edit_512.aimodel`** and
161
  **`Transformer_edit_2ref.aimodel`** for FLUX.2's in-context editing. You give a reference image and an instruction —
162
  *"add a red wizard hat, keep everything else the same"* — and only the instructed change is applied while the
163
  subject, pose, and background are preserved. This is different from strength-based image-to-image (SDEdit), which