Hugging Face's logo

mlx-community-staging
/
gemma-3-1b-it-mlx-4Bit-dynamic

Text Generation
Transformers
Safetensors
MLX
gemma3_text
conversational
text-generation-inference
2-bit
Model card Files Community
pcuenq HF Staff commited on
Commit
f9e8066
·
verified ·
1 Parent(s): ff1d870

Upload config.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. config.json +20 -20
config.json CHANGED
@@ -88,10 +88,7 @@
88
  "bits": 6,
89
  "group_size": 64
90
  },
91
- "model.layers.2.self_attn.o_proj": {
92
- "bits": 6,
93
- "group_size": 64
94
- },
95
  "model.layers.2.self_attn.q_norm": false,
96
  "model.layers.2.self_attn.k_norm": false,
97
  "model.layers.2.self_attn.rope": false,
@@ -515,7 +512,10 @@
515
  "bits": 6,
516
  "group_size": 64
517
  },
518
- "model.layers.18.self_attn.v_proj": true,
 
 
 
519
  "model.layers.18.self_attn.o_proj": true,
520
  "model.layers.18.self_attn.q_norm": false,
521
  "model.layers.18.self_attn.k_norm": false,
@@ -525,10 +525,7 @@
525
  "bits": 6,
526
  "group_size": 64
527
  },
528
- "model.layers.18.mlp.up_proj": {
529
- "bits": 6,
530
- "group_size": 64
531
- },
532
  "model.layers.18.input_layernorm": false,
533
  "model.layers.18.post_attention_layernorm": false,
534
  "model.layers.18.pre_feedforward_layernorm": false,
@@ -555,7 +552,10 @@
555
  "model.layers.19.post_feedforward_layernorm": false,
556
  "model.layers.20.self_attn.q_proj": true,
557
  "model.layers.20.self_attn.k_proj": true,
558
- "model.layers.20.self_attn.v_proj": true,
 
 
 
559
  "model.layers.20.self_attn.o_proj": {
560
  "bits": 6,
561
  "group_size": 64
@@ -750,10 +750,7 @@
750
  "bits": 6,
751
  "group_size": 64
752
  },
753
- "model.layers.2.self_attn.o_proj": {
754
- "bits": 6,
755
- "group_size": 64
756
- },
757
  "model.layers.2.self_attn.q_norm": false,
758
  "model.layers.2.self_attn.k_norm": false,
759
  "model.layers.2.self_attn.rope": false,
@@ -1177,7 +1174,10 @@
1177
  "bits": 6,
1178
  "group_size": 64
1179
  },
1180
- "model.layers.18.self_attn.v_proj": true,
 
 
 
1181
  "model.layers.18.self_attn.o_proj": true,
1182
  "model.layers.18.self_attn.q_norm": false,
1183
  "model.layers.18.self_attn.k_norm": false,
@@ -1187,10 +1187,7 @@
1187
  "bits": 6,
1188
  "group_size": 64
1189
  },
1190
- "model.layers.18.mlp.up_proj": {
1191
- "bits": 6,
1192
- "group_size": 64
1193
- },
1194
  "model.layers.18.input_layernorm": false,
1195
  "model.layers.18.post_attention_layernorm": false,
1196
  "model.layers.18.pre_feedforward_layernorm": false,
@@ -1217,7 +1214,10 @@
1217
  "model.layers.19.post_feedforward_layernorm": false,
1218
  "model.layers.20.self_attn.q_proj": true,
1219
  "model.layers.20.self_attn.k_proj": true,
1220
- "model.layers.20.self_attn.v_proj": true,
 
 
 
1221
  "model.layers.20.self_attn.o_proj": {
1222
  "bits": 6,
1223
  "group_size": 64
 
88
  "bits": 6,
89
  "group_size": 64
90
  },
91
+ "model.layers.2.self_attn.o_proj": true,
 
 
 
92
  "model.layers.2.self_attn.q_norm": false,
93
  "model.layers.2.self_attn.k_norm": false,
94
  "model.layers.2.self_attn.rope": false,
 
512
  "bits": 6,
513
  "group_size": 64
514
  },
515
+ "model.layers.18.self_attn.v_proj": {
516
+ "bits": 6,
517
+ "group_size": 64
518
+ },
519
  "model.layers.18.self_attn.o_proj": true,
520
  "model.layers.18.self_attn.q_norm": false,
521
  "model.layers.18.self_attn.k_norm": false,
 
525
  "bits": 6,
526
  "group_size": 64
527
  },
528
+ "model.layers.18.mlp.up_proj": true,
 
 
 
529
  "model.layers.18.input_layernorm": false,
530
  "model.layers.18.post_attention_layernorm": false,
531
  "model.layers.18.pre_feedforward_layernorm": false,
 
552
  "model.layers.19.post_feedforward_layernorm": false,
553
  "model.layers.20.self_attn.q_proj": true,
554
  "model.layers.20.self_attn.k_proj": true,
555
+ "model.layers.20.self_attn.v_proj": {
556
+ "bits": 6,
557
+ "group_size": 64
558
+ },
559
  "model.layers.20.self_attn.o_proj": {
560
  "bits": 6,
561
  "group_size": 64
 
750
  "bits": 6,
751
  "group_size": 64
752
  },
753
+ "model.layers.2.self_attn.o_proj": true,
 
 
 
754
  "model.layers.2.self_attn.q_norm": false,
755
  "model.layers.2.self_attn.k_norm": false,
756
  "model.layers.2.self_attn.rope": false,
 
1174
  "bits": 6,
1175
  "group_size": 64
1176
  },
1177
+ "model.layers.18.self_attn.v_proj": {
1178
+ "bits": 6,
1179
+ "group_size": 64
1180
+ },
1181
  "model.layers.18.self_attn.o_proj": true,
1182
  "model.layers.18.self_attn.q_norm": false,
1183
  "model.layers.18.self_attn.k_norm": false,
 
1187
  "bits": 6,
1188
  "group_size": 64
1189
  },
1190
+ "model.layers.18.mlp.up_proj": true,
 
 
 
1191
  "model.layers.18.input_layernorm": false,
1192
  "model.layers.18.post_attention_layernorm": false,
1193
  "model.layers.18.pre_feedforward_layernorm": false,
 
1214
  "model.layers.19.post_feedforward_layernorm": false,
1215
  "model.layers.20.self_attn.q_proj": true,
1216
  "model.layers.20.self_attn.k_proj": true,
1217
+ "model.layers.20.self_attn.v_proj": {
1218
+ "bits": 6,
1219
+ "group_size": 64
1220
+ },
1221
  "model.layers.20.self_attn.o_proj": {
1222
  "bits": 6,
1223
  "group_size": 64