Delta-Vector commited on
Commit
b70e8cc
·
verified ·
1 Parent(s): 1ac648d

Upload README.md

Browse files
Files changed (1) hide show
  1. README.md +420 -0
README.md CHANGED
@@ -0,0 +1,420 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ thumbnail: "https://cdn-uploads.huggingface.co/production/uploads/66c26b6fb01b19d8c3c2467b/jg2NWmCUfPyzizm2USjMt.jpeg"
3
+ datasets:
4
+ - NewEden/Orion-LIT
5
+ - NewEden/Orion-Asstr-Stories-16K
6
+ - Mielikki/Erebus-87k
7
+ - NewEden/Hydrus-R1-Thinking-Sharegpt
8
+ - PocketDoc/Dans-MemoryCore-CoreCurriculum-Small
9
+ - Nitral-AI/ARES-ShareGPT
10
+ - NewEden/Hydrus-HelpSteer2
11
+ - PocketDoc/Dans-Codemaxx-CodeFeedback-Conversations
12
+ - PocketDoc/Dans-Toolmaxx-Agent
13
+ - PocketDoc/Dans-Assistantmaxx-Tulu3-IF
14
+ - NewEden/Hydrus-SonnetOrca
15
+ - NewEden/Hydrus-Chat_error-Pure-Dove-sharegpt
16
+ - NewEden/No_Robots-R1-Filtered
17
+ - NewEden/GSM8K-R1-filtered
18
+ - NewEden/Hydrus_Anthropic_hh_harmful-sharegpt
19
+ - NewEden/Hydrus-Instruct-SmolTalk
20
+ - PocketDoc/Dans-Logicmaxx-Skunkworks
21
+ - PocketDoc/Dans-Logicmaxx-SAT-AP
22
+ - PocketDoc/Dans-Toolmaxx-ShellCommands
23
+ - PocketDoc/Dans-Taskmaxx-Edit
24
+ base_model:
25
+ - Qwen/QwQ
26
+ tags:
27
+ - qwen
28
+ - roleplay
29
+ - finetune
30
+ - storywriting
31
+ ---
32
+ <!DOCTYPE html>
33
+ <style>
34
+ html, body {
35
+ background: black;
36
+ color: #c9d1d9 !important;
37
+ font-family: 'Segoe UI', Tahoma, Geneva, Verdana, sans-serif;
38
+ margin: 0;
39
+ padding: 0;
40
+ min-height: 100vh;
41
+ }
42
+ .markdown-body {
43
+ color: white;
44
+ margin: 40px auto;
45
+ padding: 40px;
46
+ border-radius: 12px;
47
+ position: relative;
48
+ overflow: hidden;
49
+ }
50
+
51
+ .markdown-body::after {
52
+ content: '';
53
+ position: absolute;
54
+ top: 0;
55
+ left: 0;
56
+ width: 100%;
57
+ height: 100%;
58
+ background: #0c0f18; /* background color */
59
+ pointer-events: none;
60
+ z-index: -999;
61
+ }
62
+
63
+ h1, h2, h3 {
64
+ background: linear-gradient(45deg, #6e00ff, #00ffff);
65
+ -webkit-background-clip: text;
66
+ -webkit-text-fill-color: transparent;
67
+ border-bottom: 1px solid #333;
68
+ padding-bottom: 0.3em;
69
+ }
70
+
71
+ div[style*="border:2px solid #333"],
72
+ div[style*="border: 2px solid #333"],
73
+ div[style*="border:1px solid #333"],
74
+ div[style*="border: 1px solid #333"] {
75
+ background: rgba(22, 27, 34, 0.8) !important;
76
+ border: 2px solid #6e00ff !important;
77
+ box-shadow: 0 0 15px rgba(110, 0, 255, 0.5);
78
+ border-radius: 10px;
79
+ padding: 20px;
80
+ margin: 20px 0;
81
+ }
82
+
83
+ code {
84
+ background-color: #1a1a1a !important;
85
+ border-radius: 4px;
86
+ padding: 0.2em 0.4em;
87
+ color: #00ffff;
88
+ }
89
+
90
+ pre {
91
+ background-color: #1a1a1a !important;
92
+ border: 1px solid #333;
93
+ border-radius: 8px;
94
+ padding: 16px;
95
+ }
96
+
97
+ table {
98
+ width: 100%;
99
+ border-collapse: collapse;
100
+ margin: 20px 0;
101
+ background: rgba(0,0,0,0.2);
102
+ table-layout: fixed;
103
+ color: white;
104
+ }
105
+
106
+ th, td {
107
+ border: 1px solid #333;
108
+ padding: 12px;
109
+ text-align: center;
110
+ color: white;
111
+ }
112
+
113
+ th {
114
+ background: rgba(110, 0, 255, 0.1);
115
+ }
116
+
117
+ td:nth-child(1) {
118
+ width: 1%;
119
+ white-space: nowrap;
120
+ }
121
+
122
+ td:nth-child(2) {
123
+ width: 100%;
124
+ }
125
+
126
+ td > span {
127
+ display: block;
128
+ padding: 4px 8px;
129
+ background: rgba(110, 0, 255, 0.1);
130
+ border-radius: 4px;
131
+ transition: all 0.3s ease;
132
+ }
133
+
134
+ td > span:hover {
135
+ background: rgba(110, 0, 255, 0.2);
136
+ transform: translateY(-1px);
137
+ }
138
+
139
+ a {
140
+ color: #00ffff;
141
+ text-decoration: none;
142
+ transition: all 0.3s ease;
143
+ }
144
+
145
+ a:hover {
146
+ color: #6e00ff;
147
+ text-decoration: none;
148
+ }
149
+
150
+ hr {
151
+ border: 0;
152
+ height: 1px;
153
+ background: linear-gradient(90deg, transparent, #333, transparent);
154
+ margin: 40px 0;
155
+ }
156
+
157
+ img {
158
+ max-width: 100%;
159
+ border-radius: 10px;
160
+ }
161
+
162
+ details summary:hover {
163
+ color: #00ffff;
164
+ }
165
+
166
+ * {
167
+ color-scheme: dark !important;
168
+ }
169
+
170
+ .prose, .max-w-none, .px-4 {
171
+ background-color: transparent !important;
172
+ color: #c9d1d9 !important;
173
+ }
174
+ </style>
175
+ <body>
176
+ <div class="markdown-body">
177
+ <div align="center">
178
+
179
+ <img src="https://cdn-uploads.huggingface.co/production/uploads/66c26b6fb01b19d8c3c2467b/o5WjJKA9f95ri9UzRxZQE.png" alt="Model Visualization" width="500px" style="border: 3px solid #333; box-shadow: 0 0 15px rgba(66, 0, 131, 0.5);" />
180
+
181
+ <br>
182
+ <br>
183
+
184
+ <div style="font-size:1.5em; font-weight:bold; background: linear-gradient(45deg, #6e00ff, #00ffff); -webkit-background-clip: text; -webkit-text-fill-color: transparent;">
185
+ Hamanasu 32B
186
+ </div>
187
+
188
+ </div>
189
+
190
+ <div style="border:1px solid #333; border-radius:10px; padding:20px; margin:20px 0; background: rgba(0,0,0,0.4);">
191
+
192
+
193
+ ## 🌌 Overview
194
+
195
+ <i>This model is the Instruct tuned version of Hamanasu-QwQ-V1, This model removes the reasoning gimmick of QwQ.</i>
196
+ <i>The model has dry but good prose and stays terse, All thanks to Ruka-Hamanasu for funding the train.</i>
197
+
198
+ </div>
199
+
200
+ <div style="display: grid; grid-template-columns: repeat(auto-fit, minmax(250px, 1fr)); gap: 20px; margin: 20px 0;">
201
+
202
+ <div style="border:2px solid #333; border-radius:10px; padding:20px; background: rgba(0,0,0,0.2);">
203
+
204
+ ### 📈 Quantizations
205
+ | Type | Link |
206
+ |:---:|:---:|
207
+ | `GGUF` | <span style="display:block; width:100%">https://huggingface.co/Delta-Vector/Hamanasu-QwQ-V1.5-Instruct-gguf></span> |
208
+ | `EXL2` | <span style="display:block; width:100%">https://huggingface.co/Delta-Vector/Hamanasu-QwQ-V1.5-Instruct-EXL2</span> |
209
+
210
+ </div>
211
+
212
+ <div style="border:2px solid #333; border-radius:10px; padding:20px; background: rgba(0,0,0,0.2);">
213
+
214
+ ### ⚔️ Hardware
215
+ - 8x H100s
216
+ - Epochs: 2
217
+ - Base: `Delta-Vector/Hamanasu-32B-V1-QwQ`
218
+ - Amount of Tokens: 60M
219
+ </div>
220
+
221
+ </div>
222
+
223
+
224
+ <div style="border: 2px solid #6e00ff; border-radius: 10px; padding: 20px; margin: 20px 0; box-shadow: 0 0 15px rgba(110, 0, 255, 0.5);">
225
+
226
+ ## 💰 Prompting
227
+
228
+
229
+ <i>This model uses ChatML formatting</i>
230
+ ```python
231
+ <|im_start|>system
232
+ You are an uncensored AI, your job is to fulfill thy will of thy user.<|im_end|>
233
+ <|im_start|>User request
234
+ Take off your helmet.<|im_end|>
235
+ <|im_start|>No i shall not. This is the way.
236
+ ```
237
+ </div>
238
+
239
+ <div style="border: 2px solid #6e00ff; border-radius: 10px; padding: 20px; margin: 20px 0; box-shadow: 0 0 15px rgba(110, 0, 255, 0.5);">
240
+
241
+ ## 🎲 Recommended Sampler Preset
242
+
243
+ ```python
244
+ temperature: 1.1
245
+ min_p: 0.1
246
+ ```
247
+ </div>
248
+
249
+
250
+
251
+ <div style="border: 2px solid #6e00ff; border-radius: 10px; padding: 20px; margin: 20px 0; box-shadow: 0 0 15px rgba(110, 0, 255, 0.5);">
252
+
253
+ ## Axolotl Config ꒰(˶• ᴗ •˶)꒱
254
+
255
+ <details>
256
+
257
+ ```yaml
258
+ base_model: NewEden/Hamanasu-32B-V1
259
+ model_type: AutoModelForCausalLM
260
+ tokenizer_type: AutoTokenizer
261
+
262
+ hub_model_id: NewEden/Hamanasu-FFT-Instruct
263
+ hub_strategy: "all_checkpoints"
264
+ push_dataset_to_hub:
265
+ hf_use_auth_token: true
266
+
267
+ plugins:
268
+ - axolotl.integrations.liger.LigerPlugin
269
+ liger_rope: true
270
+ liger_rms_norm: true
271
+ liger_swiglu: true
272
+ liger_fused_linear_cross_entropy: true
273
+
274
+ load_in_8bit: false
275
+ load_in_4bit: false
276
+ strict: false
277
+
278
+ datasets:
279
+ - path: NewEden/Hydrus-R1-Thinking-Sharegpt
280
+ type: dan-chat-advanced
281
+ - path: PocketDoc/Dans-MemoryCore-CoreCurriculum-Small
282
+ type: dan-chat-advanced
283
+ - path: Nitral-AI/ARES-ShareGPT
284
+ type: dan-chat-advanced
285
+ - path: NewEden/Hydrus-HelpSteer2
286
+ type: dan-chat-advanced
287
+ - path: PocketDoc/Dans-Codemaxx-CodeFeedback-Conversations
288
+ type: dan-chat-advanced
289
+ - path: PocketDoc/Dans-Toolmaxx-Agent
290
+ type: dan-chat-advanced
291
+ - path: PocketDoc/Dans-Assistantmaxx-Tulu3-IF
292
+ type: dan-chat-advanced
293
+ - path: NewEden/Hydrus-SonnetOrca
294
+ type: dan-chat-advanced
295
+ - path: NewEden/Hydrus-Chat_error-Pure-Dove-sharegpt
296
+ type: dan-chat-advanced
297
+ - path: NewEden/No_Robots-R1-Filtered
298
+ type: dan-chat-advanced
299
+ - path: NewEden/GSM8K-R1-filtered
300
+ type: dan-chat-advanced
301
+ - path: NewEden/Hydrus_Anthropic_hh_harmful-sharegpt
302
+ type: dan-chat-advanced
303
+ - path: NewEden/Hydrus-Instruct-SmolTalk
304
+ type: dan-chat-advanced
305
+ - path: PocketDoc/Dans-Logicmaxx-Skunkworks
306
+ type: dan-chat-advanced
307
+ - path: PocketDoc/Dans-Logicmaxx-SAT-AP
308
+ type: dan-chat-advanced
309
+ - path: PocketDoc/Dans-Toolmaxx-ShellCommands
310
+ type: dan-chat-advanced
311
+ - path: PocketDoc/Dans-Taskmaxx-Edit
312
+ type: dan-chat-advanced
313
+ dataset_prepared_path: prepared_data
314
+ val_set_size: 0.0
315
+ output_dir: ./qwq-inst
316
+
317
+ sequence_len: 32768
318
+ sample_packing: true
319
+ pad_to_sequence_len: true
320
+
321
+ wandb_project: qwq
322
+ wandb_entity:
323
+ wandb_watch:
324
+ wandb_name: instruct-attempt-03
325
+ wandb_log_model:
326
+
327
+ gradient_accumulation_steps: 2
328
+ micro_batch_size: 1
329
+ num_epochs: 2
330
+ optimizer: paged_adamw_8bit
331
+ lr_scheduler: cosine
332
+ learning_rate: 5e-6
333
+
334
+ train_on_inputs: false
335
+ group_by_length: false
336
+ bf16: auto
337
+ fp16:
338
+ tf32: false
339
+
340
+ gradient_checkpointing: true
341
+ early_stopping_patience:
342
+ resume_from_checkpoint:
343
+ local_rank:
344
+ logging_steps: 1
345
+ xformers_attention:
346
+ flash_attention: true
347
+
348
+ warmup_steps: 40
349
+ evals_per_epoch:
350
+ eval_table_size:
351
+ eval_max_new_tokens:
352
+ saves_per_epoch: 2
353
+ debug:
354
+ deepspeed: deepspeed_configs/zero3_bf16.json
355
+ weight_decay: 0.02
356
+ fsdp:
357
+ fsdp_config:
358
+ special_tokens:
359
+ ```
360
+
361
+ </details>
362
+ </div>
363
+
364
+ <div align="center">
365
+
366
+ <div style="border: 2px solid #6e00ff; border-radius: 10px; padding: 20px; margin: 20px 0; box-shadow: 0 0 15px rgba(110, 0, 255, 0.5);">
367
+
368
+ ## ⚡ Credits
369
+ <div style="display: flex; justify-content: center;">
370
+ <div style="display: grid; grid-template-columns: repeat(auto-fit, minmax(200px, 1fr)); gap: 10px; margin: 20px 0; max-width: 600px;">
371
+
372
+ <div style="border:1px solid #333; padding:10px; border-radius:5px; text-align:center; background: rgba(0,0,0,0.2); display: flex; align-items: center; justify-content: center;">
373
+ <a href="https://huggingface.co/lucyknada">
374
+ <img src="https://img.shields.io/badge/%F0%9F%8C%9F-Lucy_Knada-blueviolet" alt="Lucy Knada">
375
+ </a>
376
+ </div>
377
+
378
+ <div style="border:1px solid #333; padding:10px; border-radius:5px; text-align:center; background: rgba(0,0,0,0.2); display: flex; align-items: center; justify-content: center;">
379
+ <a href="https://huggingface.co/hamanasu">
380
+ <img src="https://img.shields.io/badge/%E2%9A%94%EF%B8%8F-jeiku-blueviolet" alt="Ruka">
381
+ </a>
382
+ </div>
383
+
384
+ <div style="border:1px solid #333; padding:10px; border-radius:5px; text-align:center; background: rgba(0,0,0,0.2); display: flex; align-items: center; justify-content: center;">
385
+ <a href="https://huggingface.co/intervitens">
386
+ <img src="https://img.shields.io/badge/%F0%9F%9B%A1%EF%B8%8F-Intervitens-blueviolet" alt="Intervitens">
387
+ </a>
388
+ </div>
389
+
390
+ <div style="border:1px solid #333; padding:10px; border-radius:5px; text-align:center; background: rgba(0,0,0,0.2); display: flex; align-items: center; justify-content: center;">
391
+ <a href="https://huggingface.co/kalomaze">
392
+ <img src="https://img.shields.io/badge/%F0%9F%94%AE-Kalomaze-blueviolet" alt="Kalomaze">
393
+ </a>
394
+ </div>
395
+
396
+ <div style="border:1px solid #333; padding:10px; border-radius:5px; text-align:center; background: rgba(0,0,0,0.2); display: flex; align-items: center; justify-content: center;">
397
+ <a href="https://huggingface.co/kubernetes-bad">
398
+ <img src="https://img.shields.io/badge/%E2%9A%A1-Kubernetes_Bad-blueviolet" alt="Kubernetes Bad">
399
+ </a>
400
+ </div>
401
+
402
+ <div style="border:1px solid #333; padding:10px; border-radius:5px; text-align:center; background: rgba(0,0,0,0.2); display: flex; align-items: center; justify-content: center;">
403
+ <a href="https://huggingface.co/anthracite-org">
404
+ <img src="https://img.shields.io/badge/%F0%9F%8C%91-Anthracite-blueviolet" alt="Anthracite">
405
+ </a>
406
+ </div>
407
+ </div>
408
+ </div>
409
+ </div>
410
+
411
+ ---
412
+
413
+ <div align="center">
414
+ <div style="font-size:0.8em; opacity:0.8;">Made by</div>
415
+ <div style="font-size:1.2em; font-weight:bold; background: linear-gradient(45deg, #6e00ff, #00ffff); -webkit-background-clip: text; -webkit-text-fill-color: transparent;">Delta-Vector</div>
416
+ </div>
417
+
418
+ </div>
419
+ </body>
420
+ </html>