Hugging Face's logo

mlboydaisuke
/
gemma-4-E4B-CoreAI

Text Generation
coreai-aimodel
coreai
aimodel
apple-silicon
on-device
gemma-4
qat
gpu-pipelined
Model card Files
xet
 Community
mlboydaisuke commited on
Commit
7944076
·
verified ·
1 Parent(s): f6e31d9

E4B QAT gather tables (PLE int8 per-row dump + scales)

Browse files
Files changed (8) hide show
  1. .gitattributes +5 -0
  2. ios-frontend/gemma4_e4b_qat_gather_raw/embed_per_layer.i8 +3 -0
  3. ios-frontend/gemma4_e4b_qat_gather_raw/embed_per_layer.scale.f32 +3 -0
  4. ios-frontend/gemma4_e4b_qat_gather_raw/embed_tokens.i8 +3 -0
  5. ios-frontend/gemma4_e4b_qat_gather_raw/embed_tokens.scale.f32 +3 -0
  6. ios-frontend/gemma4_e4b_qat_gather_raw/meta.json +12 -0
  7. ios-frontend/gemma4_e4b_qat_gather_raw/proj.f32 +3 -0
  8. ios-frontend/gemma4_e4b_qat_gather_raw/proj_norm.f32 +0 -0
.gitattributes CHANGED
@@ -40,3 +40,8 @@ gpu-pipelined/gemma4_e4b_qat_decode_int4lin_aotc_h18p/gemma4_e4b_qat_decode_int4
40
  gpu-pipelined/gemma4_e4b_qat_decode_int4lin_aotc_h18p/tokenizer/tokenizer.json filter=lfs diff=lfs merge=lfs -text
41
  gpu-pipelined/gemma4_e4b_qat_decode_int4lin_tbl/gemma4_e4b_qat_decode_int4lin_tbl.aimodel/main.mlirb filter=lfs diff=lfs merge=lfs -text
42
  gpu-pipelined/gemma4_e4b_qat_decode_int4lin_tbl/tokenizer/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 
 
 
 
 
 
40
  gpu-pipelined/gemma4_e4b_qat_decode_int4lin_aotc_h18p/tokenizer/tokenizer.json filter=lfs diff=lfs merge=lfs -text
41
  gpu-pipelined/gemma4_e4b_qat_decode_int4lin_tbl/gemma4_e4b_qat_decode_int4lin_tbl.aimodel/main.mlirb filter=lfs diff=lfs merge=lfs -text
42
  gpu-pipelined/gemma4_e4b_qat_decode_int4lin_tbl/tokenizer/tokenizer.json filter=lfs diff=lfs merge=lfs -text
43
+ ios-frontend/gemma4_e4b_qat_gather_raw/embed_per_layer.i8 filter=lfs diff=lfs merge=lfs -text
44
+ ios-frontend/gemma4_e4b_qat_gather_raw/embed_per_layer.scale.f32 filter=lfs diff=lfs merge=lfs -text
45
+ ios-frontend/gemma4_e4b_qat_gather_raw/embed_tokens.i8 filter=lfs diff=lfs merge=lfs -text
46
+ ios-frontend/gemma4_e4b_qat_gather_raw/embed_tokens.scale.f32 filter=lfs diff=lfs merge=lfs -text
47
+ ios-frontend/gemma4_e4b_qat_gather_raw/proj.f32 filter=lfs diff=lfs merge=lfs -text
ios-frontend/gemma4_e4b_qat_gather_raw/embed_per_layer.i8 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:891f77177b4ab317e5562b6684c5f7825cd5795a9c4b26f97af121ebf4bb7ccb
3
+ size 2818572288
ios-frontend/gemma4_e4b_qat_gather_raw/embed_per_layer.scale.f32 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:512d9d0a2b089ae37594ce80e848a08e1b5a6b3f40014a32333e6d191a9b6131
3
+ size 1048576
ios-frontend/gemma4_e4b_qat_gather_raw/embed_tokens.i8 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:035fd90b534b1c71a50e7d2a5524cda1b3492dd07b78923c1b7fd7f29f6eff7e
3
+ size 671088640
ios-frontend/gemma4_e4b_qat_gather_raw/embed_tokens.scale.f32 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:11e4a961bb1d30b20ad09a1260b088093312d516453a0e4d5318af470d6575ff
3
+ size 1048576
ios-frontend/gemma4_e4b_qat_gather_raw/meta.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "V": 262144,
3
+ "D": 2560,
4
+ "PLD": 10752,
5
+ "L": 42,
6
+ "ld": 256,
7
+ "embed_scale_main": 50.59644256269407,
8
+ "embed_scale_pl": 16.0,
9
+ "proj_scale": 0.01976423537605237,
10
+ "input_scale": 0.7071067811865476,
11
+ "rms_eps": 1e-06
12
+ }
ios-frontend/gemma4_e4b_qat_gather_raw/proj.f32 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:529031a1a2774f93a85018ef711e382e004b3e35c9e551a490d2f032e5adbd98
3
+ size 110100480
ios-frontend/gemma4_e4b_qat_gather_raw/proj_norm.f32 ADDED
Binary file (1.02 kB). View file