Hugging Face's logo

rty813
/
humanizer-bf16

Text Generation
Transformers
PyTorch
gemma4_unified
image-text-to-text
bf16
humanizer
full-precision
vllm
xinference
Model card Files
xet
 Community
rty813 commited on
Commit
082fe30
·
verified ·
1 Parent(s): 84aab13

Add humanizer-bf16 weights (bf16)

Browse files
Files changed (3) hide show
  1. README.md +60 -0
  2. config.json +166 -0
  3. pytorch_model.bin +3 -0
README.md ADDED
@@ -0,0 +1,60 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ base_model: jialinyyzz/humanizer
4
+ pipeline_tag: text-generation
5
+ library_name: transformers
6
+ tags:
7
+ - bf16
8
+ - humanizer
9
+ - full-precision
10
+ - transformers
11
+ - vllm
12
+ - xinference
13
+ ---
14
+
15
+ # humanizer-bf16 (bf16)
16
+
17
+ Full-precision (bf16) PyTorch port of [jialinyyzz/humanizer](https://huggingface.co/jialinyyzz/humanizer),
18
+ for vLLM / TGI / Xinference / transformers users who don't want the GGUF quant.
19
+
20
+ - Format: `pytorch_model.bin` (bf16). safetensors conversion in progress.
21
+ - Architecture: identical to the original — drop-in for `AutoModelForCausalLM.from_pretrained("rty813/humanizer-bf16")`.
22
+ - Also loads in vLLM (`vllm serve rty813/humanizer-bf16`) and Xinference (search "humanizer").
23
+
24
+ Original model card:
25
+
26
+ ---
27
+ license: apache-2.0
28
+ base_model: google/gemma-4-12B
29
+ base_model_relation: finetune
30
+ language:
31
+ - en
32
+ - zh
33
+ pipeline_tag: text-generation
34
+ library_name: gguf
35
+ tags:
36
+ - humanizer
37
+ - transformers
38
+ - safetensors
39
+ - text-rewriting
40
+ - rewriting
41
+ - paraphrase
42
+ - style-transfer
43
+ - gguf
44
+ - llama.cpp
45
+ - gemma4
46
+ ---
47
+
48
+ <img src="assets/banner-en.png" alt="humanizer: rewrites AI drafts so they read like a person wrote them" width="100%">
49
+
50
+ <img src="assets/app-showcase-en.png" alt="The humanizer app running the 12B model locally: draft on the left, rewrite on the right, new wording highlighted" width="100%">
51
+
52
+ # humanizer
53
+
54
+ **A 12B model that rewrites AI-written drafts (emails, essays, reports, forum posts; English and Chinese) so they read like a person wrote them.** It is trained to keep every number, unit, date, name and quote and to add nothing, and it runs locally.
55
+
56
+ **Quick start:** get the app for **[Mac (Apple silicon)](https://github.com/sgaofen/humanizer-local-model/releases/download/app-v0.3.2/Humanizer-0.3.2-macos-arm64.dmg)** or **[Windows](https://github.com/sgaofen/humanizer-local-model/releases/download/app-v0.3.2/Humanizer-0.3.2-windows-x64-setup.exe)**, open it and paste a draft; it works offline. Without the app: [llama.cpp and other runtimes](#usage-without-the-app), or [AGENTS.md](AGENTS.md) for coding agents.
57
+
58
+ ## Before and after
59
+
60
+ Drafts from the held-out evaluation set. For each we generated 8 samples with the Q8_0 file (llama.cpp, the app's settings) and picked the one that reads best
config.json ADDED
@@ -0,0 +1,166 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Gemma4UnifiedForConditionalGeneration"
4
+ ],
5
+ "audio_config": {
6
+ "_name_or_path": "",
7
+ "architectures": null,
8
+ "audio_embed_dim": 640,
9
+ "chunk_size_feed_forward": 0,
10
+ "dtype": "bfloat16",
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1"
14
+ },
15
+ "initializer_range": 0.02,
16
+ "is_encoder_decoder": false,
17
+ "label2id": {
18
+ "LABEL_0": 0,
19
+ "LABEL_1": 1
20
+ },
21
+ "model_type": "gemma4_unified_audio",
22
+ "output_attentions": false,
23
+ "output_hidden_states": false,
24
+ "problem_type": null,
25
+ "return_dict": true,
26
+ "rms_norm_eps": 1e-06
27
+ },
28
+ "audio_token_id": 258881,
29
+ "boa_token_id": 256000,
30
+ "boi_token_id": 255999,
31
+ "dtype": "bfloat16",
32
+ "eoa_token_index": 258883,
33
+ "eoi_token_id": 258882,
34
+ "image_token_id": 258880,
35
+ "initializer_range": 0.02,
36
+ "model_type": "gemma4_unified",
37
+ "text_config": {
38
+ "attention_bias": false,
39
+ "attention_dropout": 0.0,
40
+ "attention_k_eq_v": true,
41
+ "bos_token_id": 2,
42
+ "dtype": "bfloat16",
43
+ "enable_moe_block": false,
44
+ "eos_token_id": 1,
45
+ "final_logit_softcapping": 30.0,
46
+ "global_head_dim": 512,
47
+ "head_dim": 256,
48
+ "hidden_activation": "gelu_pytorch_tanh",
49
+ "hidden_size": 3840,
50
+ "hidden_size_per_layer_input": 0,
51
+ "initializer_range": 0.02,
52
+ "intermediate_size": 15360,
53
+ "layer_types": [
54
+ "sliding_attention",
55
+ "sliding_attention",
56
+ "sliding_attention",
57
+ "sliding_attention",
58
+ "sliding_attention",
59
+ "full_attention",
60
+ "sliding_attention",
61
+ "sliding_attention",
62
+ "sliding_attention",
63
+ "sliding_attention",
64
+ "sliding_attention",
65
+ "full_attention",
66
+ "sliding_attention",
67
+ "sliding_attention",
68
+ "sliding_attention",
69
+ "sliding_attention",
70
+ "sliding_attention",
71
+ "full_attention",
72
+ "sliding_attention",
73
+ "sliding_attention",
74
+ "sliding_attention",
75
+ "sliding_attention",
76
+ "sliding_attention",
77
+ "full_attention",
78
+ "sliding_attention",
79
+ "sliding_attention",
80
+ "sliding_attention",
81
+ "sliding_attention",
82
+ "sliding_attention",
83
+ "full_attention",
84
+ "sliding_attention",
85
+ "sliding_attention",
86
+ "sliding_attention",
87
+ "sliding_attention",
88
+ "sliding_attention",
89
+ "full_attention",
90
+ "sliding_attention",
91
+ "sliding_attention",
92
+ "sliding_attention",
93
+ "sliding_attention",
94
+ "sliding_attention",
95
+ "full_attention",
96
+ "sliding_attention",
97
+ "sliding_attention",
98
+ "sliding_attention",
99
+ "sliding_attention",
100
+ "sliding_attention",
101
+ "full_attention"
102
+ ],
103
+ "max_position_embeddings": 262144,
104
+ "model_type": "gemma4_unified_text",
105
+ "moe_intermediate_size": null,
106
+ "num_attention_heads": 16,
107
+ "num_experts": null,
108
+ "num_global_key_value_heads": 1,
109
+ "num_hidden_layers": 48,
110
+ "num_key_value_heads": 8,
111
+ "num_kv_shared_layers": 0,
112
+ "pad_token_id": 0,
113
+ "rms_norm_eps": 1e-06,
114
+ "rope_parameters": {
115
+ "full_attention": {
116
+ "partial_rotary_factor": 0.25,
117
+ "rope_theta": 1000000.0,
118
+ "rope_type": "proportional"
119
+ },
120
+ "sliding_attention": {
121
+ "rope_theta": 10000.0,
122
+ "rope_type": "default"
123
+ }
124
+ },
125
+ "sliding_window": 1024,
126
+ "tie_word_embeddings": true,
127
+ "top_k_experts": null,
128
+ "use_bidirectional_attention": "vision",
129
+ "use_cache": true,
130
+ "use_double_wide_mlp": false,
131
+ "vocab_size": 262144,
132
+ "vocab_size_per_layer_input": 262144
133
+ },
134
+ "tie_word_embeddings": true,
135
+ "transformers_version": "5.14.1",
136
+ "video_token_id": 258884,
137
+ "vision_config": {
138
+ "_name_or_path": "",
139
+ "architectures": null,
140
+ "chunk_size_feed_forward": 0,
141
+ "dtype": "bfloat16",
142
+ "id2label": {
143
+ "0": "LABEL_0",
144
+ "1": "LABEL_1"
145
+ },
146
+ "initializer_range": 0.02,
147
+ "is_encoder_decoder": false,
148
+ "label2id": {
149
+ "LABEL_0": 0,
150
+ "LABEL_1": 1
151
+ },
152
+ "mm_embed_dim": 3840,
153
+ "mm_posemb_size": 1120,
154
+ "model_type": "gemma4_unified_vision",
155
+ "num_soft_tokens": 280,
156
+ "output_attentions": false,
157
+ "output_hidden_states": false,
158
+ "output_proj_dims": 3840,
159
+ "patch_size": 16,
160
+ "pooling_kernel_size": 3,
161
+ "problem_type": null,
162
+ "return_dict": true,
163
+ "rms_norm_eps": 1e-06
164
+ },
165
+ "torch_dtype": "bfloat16"
166
+ }
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9d2d3b413129da2c8dddfbc531b56ff6859f7b5b14fc3f0eb95bf1d38f2e6bcf
3
+ size 10240