ArtusDev commited on
Commit
9b0b155
·
verified ·
1 Parent(s): 5823756

Upload folder using huggingface_hub

Browse files
.gitattributes CHANGED
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ tokenizer.json filter=lfs diff=lfs merge=lfs -text
README.md ADDED
@@ -0,0 +1,86 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: Qwen/Qwen3-8B-Base
3
+ library_name: transformers
4
+ tags:
5
+ - mergekit
6
+ - axolotl
7
+ - unsloth
8
+ - roleplay
9
+ - conversational
10
+ datasets:
11
+ - PygmalionAI/PIPPA
12
+ - Alfitaria/nemotron-ultra-reasoning-synthkink
13
+ - PocketDoc/Dans-Prosemaxx-Gutenberg
14
+ - FreedomIntelligence/Medical-R1-Distill-Data
15
+ - cognitivecomputations/SystemChat-2.0
16
+ - allenai/tulu-3-sft-personas-instruction-following
17
+ - kalomaze/Opus_Instruct_25k
18
+ - simplescaling/s1K-claude-3-7-sonnet
19
+ - ai2-adapt-dev/flan_v2_converted
20
+ - grimulkan/theory-of-mind
21
+ - grimulkan/physical-reasoning
22
+ - nvidia/HelpSteer3
23
+ - nbeerbower/gutenberg2-dpo
24
+ - nbeerbower/gutenberg-moderne-dpo
25
+ - nbeerbower/Purpura-DPO
26
+ - antiven0m/physical-reasoning-dpo
27
+ - allenai/tulu-3-IF-augmented-on-policy-70b
28
+ - NobodyExistsOnTheInternet/system-message-DPO
29
+ ---
30
+
31
+ # Q3-8B-Kintsugi
32
+
33
+ FIXME: we need a new image
34
+
35
+ # Overview
36
+
37
+ ***Q3-8B-Kintsugi*** is a roleplaying model finetuned from [Qwen3-8B-Base](https://huggingface.co/Qwen/Qwen3-8B-Base).
38
+
39
+ During testing, Designant punched well above its weight class in terms of parameters, especially for 1-on-1 roleplaying and general storywriting.
40
+
41
+ # Quantizations
42
+
43
+
44
+ EXL3:
45
+ - TODO!
46
+
47
+ MLX:
48
+ - TODO!
49
+
50
+ GGUF:
51
+ - TODO!
52
+
53
+ # Usage
54
+
55
+ - Format is plain-old ChatML (please note that, unlike regular Qwen 3, you do *not* need to prefill empty think tags for it not to reason -- see below).
56
+
57
+ - Settings used by testers varied, but Fizz and inflatebot used the same settings and system prompt recommended for [GLM4-32B-Neon-v2.](https://huggingface.co/allura-org/GLM4-32B-Neon-v2)
58
+
59
+ - The official instruction following version of Qwen3-8B was not used as a base. Instruction-following is trained in post-hoc, and "thinking" traces were not included. __As a result of this, "thinking" will not function as intended.__
60
+
61
+ # Training Process
62
+
63
+ 1. The [base model](https://huggingface.co/Qwen/Qwen3-8B-Base) first went through a supervised finetune on a corpus of instruction following data, roleplay conversations, and human writing based on the [Ink](https://huggingface.co/collections/allura-org/ink-6772fd1442308781594bbabb)/[Bigger Body](https://huggingface.co/collections/allura-org/bigger-body-67b277af0861cec33b54745d)/[Remnant](https://huggingface.co/collections/allura-org/remnant-6817c2113bbb2aed501513d0) lineage.
64
+
65
+ 2. Finally, a KTO reinforcement learning phase steered the model away from the very purple prose the initial merge had, and improved its logical+spatial reasoning and sense of overall "intelligence".
66
+
67
+ Both stages here are very similar to [Q3-30B-A3B-Designant](https://huggingface.co/allura-org/Q3-30B-A3B-Designant), which went through a very similar process with the same data.
68
+
69
+ # Credits
70
+
71
+ - Fizz - Training, Data Wrangling
72
+
73
+ - Toaster, Mango, Bot, probably others I forgot ;-; - Testing
74
+
75
+ - inflatebot - original Designant model card that this one was yoinked from
76
+
77
+ - Artus - Funding
78
+
79
+ - Alibaba - Making the original model
80
+
81
+ - Axolotl, Unsloth, Huggingface - Making the frameworks used to train this model (Axolotl was used for the SFT process, and Unsloth+TRL was used for the KTO process)
82
+
83
+ We would like to thank the Allura community on Discord, especially Curse, Heni, Artus and Mawnipulator, for their companionship and moral support. You all mean the world to us <3
84
+
85
+ ---
86
+ <sup>*`There, God is not.`*</sup>
config.json ADDED
@@ -0,0 +1,41 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Qwen3ForCausalLM"
4
+ ],
5
+ "attention_bias": false,
6
+ "attention_dropout": 0.0,
7
+ "bos_token_id": 151643,
8
+ "eos_token_id": 151643,
9
+ "head_dim": 128,
10
+ "hidden_act": "silu",
11
+ "hidden_size": 4096,
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 12288,
14
+ "max_position_embeddings": 32768,
15
+ "max_window_layers": 36,
16
+ "model_type": "qwen3",
17
+ "num_attention_heads": 32,
18
+ "num_hidden_layers": 36,
19
+ "num_key_value_heads": 8,
20
+ "rms_norm_eps": 1e-06,
21
+ "rope_scaling": null,
22
+ "rope_theta": 1000000,
23
+ "sliding_window": null,
24
+ "tie_word_embeddings": false,
25
+ "torch_dtype": "bfloat16",
26
+ "transformers_version": "4.51.0",
27
+ "use_cache": true,
28
+ "use_sliding_window": false,
29
+ "vocab_size": 151936,
30
+ "quantization_config": {
31
+ "quant_method": "exl3",
32
+ "version": "0.0.3",
33
+ "bits": 4.5,
34
+ "head_bits": 6,
35
+ "calibration": {
36
+ "rows": 100,
37
+ "cols": 2048
38
+ },
39
+ "out_scales": "auto"
40
+ }
41
+ }
generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token_id": 151643,
3
+ "do_sample": false,
4
+ "eos_token_id": 151643,
5
+ "max_new_tokens": 2048,
6
+ "transformers_version": "4.37.0"
7
+ }
merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d9a760bdd0e9af53be65965b8a2caa1fa064f52cb9b90ed19b68ad340d6296e9
3
+ size 5624885936
quantization_config.json ADDED
The diff for this file is too large to render. See raw diff
 
special_tokens_map.json ADDED
@@ -0,0 +1,16 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "eos_token": {
3
+ "content": "<|im_end|>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "pad_token": {
10
+ "content": "<|endoftext|>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ }
16
+ }
tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:aeb13307a71acd8fe81861d94ad54ab689df773318809eed3cbe794b4492dae4
3
+ size 11422654
tokenizer_config.json ADDED
@@ -0,0 +1,220 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "added_tokens_decoder": {
3
+ "151643": {
4
+ "content": "<|endoftext|>",
5
+ "lstrip": false,
6
+ "normalized": false,
7
+ "rstrip": false,
8
+ "single_word": false,
9
+ "special": true
10
+ },
11
+ "151644": {
12
+ "content": "<|im_start|>",
13
+ "lstrip": false,
14
+ "normalized": false,
15
+ "rstrip": false,
16
+ "single_word": false,
17
+ "special": true
18
+ },
19
+ "151645": {
20
+ "content": "<|im_end|>",
21
+ "lstrip": false,
22
+ "normalized": false,
23
+ "rstrip": false,
24
+ "single_word": false,
25
+ "special": true
26
+ },
27
+ "151646": {
28
+ "content": "<|object_ref_start|>",
29
+ "lstrip": false,
30
+ "normalized": false,
31
+ "rstrip": false,
32
+ "single_word": false,
33
+ "special": true
34
+ },
35
+ "151647": {
36
+ "content": "<|object_ref_end|>",
37
+ "lstrip": false,
38
+ "normalized": false,
39
+ "rstrip": false,
40
+ "single_word": false,
41
+ "special": true
42
+ },
43
+ "151648": {
44
+ "content": "<|box_start|>",
45
+ "lstrip": false,
46
+ "normalized": false,
47
+ "rstrip": false,
48
+ "single_word": false,
49
+ "special": true
50
+ },
51
+ "151649": {
52
+ "content": "<|box_end|>",
53
+ "lstrip": false,
54
+ "normalized": false,
55
+ "rstrip": false,
56
+ "single_word": false,
57
+ "special": true
58
+ },
59
+ "151650": {
60
+ "content": "<|quad_start|>",
61
+ "lstrip": false,
62
+ "normalized": false,
63
+ "rstrip": false,
64
+ "single_word": false,
65
+ "special": true
66
+ },
67
+ "151651": {
68
+ "content": "<|quad_end|>",
69
+ "lstrip": false,
70
+ "normalized": false,
71
+ "rstrip": false,
72
+ "single_word": false,
73
+ "special": true
74
+ },
75
+ "151652": {
76
+ "content": "<|vision_start|>",
77
+ "lstrip": false,
78
+ "normalized": false,
79
+ "rstrip": false,
80
+ "single_word": false,
81
+ "special": true
82
+ },
83
+ "151653": {
84
+ "content": "<|vision_end|>",
85
+ "lstrip": false,
86
+ "normalized": false,
87
+ "rstrip": false,
88
+ "single_word": false,
89
+ "special": true
90
+ },
91
+ "151654": {
92
+ "content": "<|vision_pad|>",
93
+ "lstrip": false,
94
+ "normalized": false,
95
+ "rstrip": false,
96
+ "single_word": false,
97
+ "special": true
98
+ },
99
+ "151655": {
100
+ "content": "<|image_pad|>",
101
+ "lstrip": false,
102
+ "normalized": false,
103
+ "rstrip": false,
104
+ "single_word": false,
105
+ "special": true
106
+ },
107
+ "151656": {
108
+ "content": "<|video_pad|>",
109
+ "lstrip": false,
110
+ "normalized": false,
111
+ "rstrip": false,
112
+ "single_word": false,
113
+ "special": true
114
+ },
115
+ "151657": {
116
+ "content": "<tool_call>",
117
+ "lstrip": false,
118
+ "normalized": false,
119
+ "rstrip": false,
120
+ "single_word": false,
121
+ "special": false
122
+ },
123
+ "151658": {
124
+ "content": "</tool_call>",
125
+ "lstrip": false,
126
+ "normalized": false,
127
+ "rstrip": false,
128
+ "single_word": false,
129
+ "special": false
130
+ },
131
+ "151659": {
132
+ "content": "<|fim_prefix|>",
133
+ "lstrip": false,
134
+ "normalized": false,
135
+ "rstrip": false,
136
+ "single_word": false,
137
+ "special": false
138
+ },
139
+ "151660": {
140
+ "content": "<|fim_middle|>",
141
+ "lstrip": false,
142
+ "normalized": false,
143
+ "rstrip": false,
144
+ "single_word": false,
145
+ "special": false
146
+ },
147
+ "151661": {
148
+ "content": "<|fim_suffix|>",
149
+ "lstrip": false,
150
+ "normalized": false,
151
+ "rstrip": false,
152
+ "single_word": false,
153
+ "special": false
154
+ },
155
+ "151662": {
156
+ "content": "<|fim_pad|>",
157
+ "lstrip": false,
158
+ "normalized": false,
159
+ "rstrip": false,
160
+ "single_word": false,
161
+ "special": false
162
+ },
163
+ "151663": {
164
+ "content": "<|repo_name|>",
165
+ "lstrip": false,
166
+ "normalized": false,
167
+ "rstrip": false,
168
+ "single_word": false,
169
+ "special": false
170
+ },
171
+ "151664": {
172
+ "content": "<|file_sep|>",
173
+ "lstrip": false,
174
+ "normalized": false,
175
+ "rstrip": false,
176
+ "single_word": false,
177
+ "special": false
178
+ },
179
+ "151665": {
180
+ "content": "<tool_response>",
181
+ "lstrip": false,
182
+ "normalized": false,
183
+ "rstrip": false,
184
+ "single_word": false,
185
+ "special": false
186
+ },
187
+ "151666": {
188
+ "content": "</tool_response>",
189
+ "lstrip": false,
190
+ "normalized": false,
191
+ "rstrip": false,
192
+ "single_word": false,
193
+ "special": false
194
+ },
195
+ "151667": {
196
+ "content": "<think>",
197
+ "lstrip": false,
198
+ "normalized": false,
199
+ "rstrip": false,
200
+ "single_word": false,
201
+ "special": false
202
+ },
203
+ "151668": {
204
+ "content": "</think>",
205
+ "lstrip": false,
206
+ "normalized": false,
207
+ "rstrip": false,
208
+ "single_word": false,
209
+ "special": false
210
+ }
211
+ },
212
+ "bos_token": null,
213
+ "clean_up_tokenization_spaces": false,
214
+ "eos_token": "<|im_end|>",
215
+ "extra_special_tokens": {},
216
+ "model_max_length": 1000000000000000019884624838656,
217
+ "pad_token": "<|endoftext|>",
218
+ "tokenizer_class": "Qwen2Tokenizer",
219
+ "unk_token": null
220
+ }
vocab.json ADDED
The diff for this file is too large to render. See raw diff