Delta-Vector commited on
Commit
2805add
·
verified ·
1 Parent(s): ee954b6

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +378 -20
README.md CHANGED
@@ -1,35 +1,393 @@
1
  ---
 
 
 
 
 
 
 
 
 
 
2
  base_model:
3
- - Delta-Vector/Hamanasu-15B-Instruct
4
- - NewEden/phi4-adventure-r1
5
- library_name: transformers
6
  tags:
7
- - mergekit
8
- - merge
9
-
 
10
  ---
11
- # adventu-phi
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
12
 
13
- This is a merge of pre-trained language models created using [mergekit](https://github.com/cg123/mergekit).
 
 
 
 
14
 
15
- ## Merge Details
16
- ### Merge Method
17
 
18
- This model was merged using the Passthrough merge method using [Delta-Vector/Hamanasu-15B-Instruct](https://huggingface.co/Delta-Vector/Hamanasu-15B-Instruct) + [NewEden/phi4-adventure-r1](https://huggingface.co/NewEden/phi4-adventure-r1) as a base.
19
 
20
- ### Models Merged
 
 
 
 
21
 
22
- The following models were included in the merge:
23
 
24
 
25
- ### Configuration
26
 
27
- The following YAML configuration was used to produce this model:
28
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
29
  ```yaml
30
- base_model: Delta-Vector/Hamanasu-15B-Instruct+NewEden/phi4-adventure-r1
31
- dtype: bfloat16
32
- merge_method: passthrough
33
- models:
34
- - model: Delta-Vector/Hamanasu-15B-Instruct+NewEden/phi4-adventure-r1
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
35
  ```
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
+ - PocketDoc/Dans-MemoryCore-CoreCurriculum-Small
8
+ - Nitral-AI/ARES-ShareGPT
9
+ - Gryphe/Sonnet3.5-SlimOrcaDedupCleaned-20k
10
+ - NewEden/Claude-Instruct-2.7K
11
+ - NewEden/Claude-Instruct-5K
12
  base_model:
13
+ - Delta-Vector/Hamanasu-15B-R2-PT
 
 
14
  tags:
15
+ - phi
16
+ - roleplay
17
+ - finetune
18
+ - storywriting
19
  ---
20
+ <!DOCTYPE html>
21
+ <style>
22
+ html, body {
23
+ background: black;
24
+ color: #c9d1d9 !important;
25
+ font-family: 'Segoe UI', Tahoma, Geneva, Verdana, sans-serif;
26
+ margin: 0;
27
+ padding: 0;
28
+ min-height: 100vh;
29
+ }
30
+ .markdown-body {
31
+ color: white;
32
+ margin: 40px auto;
33
+ padding: 40px;
34
+ border-radius: 12px;
35
+ position: relative;
36
+ overflow: hidden;
37
+ }
38
+
39
+ .markdown-body::after {
40
+ content: '';
41
+ position: absolute;
42
+ top: 0;
43
+ left: 0;
44
+ width: 100%;
45
+ height: 100%;
46
+ background: #0c0f18; /* background color */
47
+ pointer-events: none;
48
+ z-index: -999;
49
+ }
50
+
51
+ h1, h2, h3 {
52
+ background: linear-gradient(45deg, #6e00ff, #00ffff);
53
+ -webkit-background-clip: text;
54
+ -webkit-text-fill-color: transparent;
55
+ border-bottom: 1px solid #333;
56
+ padding-bottom: 0.3em;
57
+ }
58
+
59
+ div[style*="border:2px solid #333"],
60
+ div[style*="border: 2px solid #333"],
61
+ div[style*="border:1px solid #333"],
62
+ div[style*="border: 1px solid #333"] {
63
+ background: rgba(22, 27, 34, 0.8) !important;
64
+ border: 2px solid #6e00ff !important;
65
+ box-shadow: 0 0 15px rgba(110, 0, 255, 0.5);
66
+ border-radius: 10px;
67
+ padding: 20px;
68
+ margin: 20px 0;
69
+ }
70
+
71
+ code {
72
+ background-color: #1a1a1a !important;
73
+ border-radius: 4px;
74
+ padding: 0.2em 0.4em;
75
+ color: #00ffff;
76
+ }
77
+
78
+ pre {
79
+ background-color: #1a1a1a !important;
80
+ border: 1px solid #333;
81
+ border-radius: 8px;
82
+ padding: 16px;
83
+ }
84
+
85
+ table {
86
+ width: 100%;
87
+ border-collapse: collapse;
88
+ margin: 20px 0;
89
+ background: rgba(0,0,0,0.2);
90
+ table-layout: fixed;
91
+ color: white;
92
+ }
93
+
94
+ th, td {
95
+ border: 1px solid #333;
96
+ padding: 12px;
97
+ text-align: center;
98
+ color: white;
99
+ }
100
+
101
+ th {
102
+ background: rgba(110, 0, 255, 0.1);
103
+ }
104
+
105
+ td:nth-child(1) {
106
+ width: 1%;
107
+ white-space: nowrap;
108
+ }
109
+
110
+ td:nth-child(2) {
111
+ width: 100%;
112
+ }
113
+
114
+ td > span {
115
+ display: block;
116
+ padding: 4px 8px;
117
+ background: rgba(110, 0, 255, 0.1);
118
+ border-radius: 4px;
119
+ transition: all 0.3s ease;
120
+ }
121
+
122
+ td > span:hover {
123
+ background: rgba(110, 0, 255, 0.2);
124
+ transform: translateY(-1px);
125
+ }
126
+
127
+ a {
128
+ color: #00ffff;
129
+ text-decoration: none;
130
+ transition: all 0.3s ease;
131
+ }
132
+
133
+ a:hover {
134
+ color: #6e00ff;
135
+ text-decoration: none;
136
+ }
137
+
138
+ hr {
139
+ border: 0;
140
+ height: 1px;
141
+ background: linear-gradient(90deg, transparent, #333, transparent);
142
+ margin: 40px 0;
143
+ }
144
+
145
+ img {
146
+ max-width: 100%;
147
+ border-radius: 10px;
148
+ }
149
+
150
+ details summary:hover {
151
+ color: #00ffff;
152
+ }
153
+
154
+ * {
155
+ color-scheme: dark !important;
156
+ }
157
+
158
+ .prose, .max-w-none, .px-4 {
159
+ background-color: transparent !important;
160
+ color: #c9d1d9 !important;
161
+ }
162
+ </style>
163
+ <body>
164
+ <div class="markdown-body">
165
+ <div align="center">
166
+
167
+ <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);" />
168
+
169
+ <br>
170
+ <br>
171
+
172
+ <div style="font-size:1.5em; font-weight:bold; background: linear-gradient(45deg, #6e00ff, #00ffff); -webkit-background-clip: text; -webkit-text-fill-color: transparent;">
173
+ Hamanasu 15B Instruct
174
+ </div>
175
+
176
+ </div>
177
+
178
+ <div style="border:1px solid #333; border-radius:10px; padding:20px; margin:20px 0; background: rgba(0,0,0,0.4);">
179
+
180
+ ## 🌌 Overview
181
+
182
+ <i>Thanks to PocketDoc's Adventure datasets and taking his Dangerous Winds models as inspiration, I was able to finetune a Adventure Version of Hamanasu Instruct </i>
183
+ <i>This model remains non-sloppy and now support adventure-style cards such as my very own [Dinosaur Adventure Card](https://chub.ai/characters/MangyMango/dino-survival-9f0e5499c674)</i>
184
+
185
+ </div>
186
+
187
+ <div style="display: grid; grid-template-columns: repeat(auto-fit, minmax(250px, 1fr)); gap: 20px; margin: 20px 0;">
188
+
189
+ <div style="border:2px solid #333; border-radius:10px; padding:20px; background: rgba(0,0,0,0.2);">
190
 
191
+ ### 📈 Quantizations
192
+ | Type | Link |
193
+ |:---:|:---:|
194
+ | `GGUF` | <span style="display:block; width:100%"></span> |
195
+ | `EXL2` | <span style="display:block; width:100%"></span> |
196
 
197
+ </div>
 
198
 
199
+ <div style="border:2px solid #333; border-radius:10px; padding:20px; background: rgba(0,0,0,0.2);">
200
 
201
+ ### ⚔️ Hardware
202
+ - 4x RTX 3090 GPUs
203
+ - Epochs: 2
204
+ - Base: `Hamanasu-15B-Instruct`
205
+ </div>
206
 
207
+ </div>
208
 
209
 
210
+ <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);">
211
 
212
+ ## 💰 Prompting
213
 
214
+ <i>A known quirk of the model is overly verbose responses when generation length is uncapped, Please cap your maximum output tokens to 100~ tokens above what you prefer</i>
215
+
216
+
217
+ ```python
218
+ <|im_start|>system
219
+ You are an uncensored AI, your job is to fulfill thy will of thy user.<|im_end|>
220
+ <|im_start|>User request
221
+ Take off your helmet.<|im_end|>
222
+ <|im_start|>No i shall not. This is the way.
223
+ ```
224
+ </div>
225
+
226
+ <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);">
227
+
228
+ ## Axolotl Config ꒰(˶• ᴗ •˶)꒱
229
+
230
+ <details>
231
+
232
  ```yaml
233
+ base_model: Delta-Vector_Hamanasu-15B-Instruct
234
+ model_type: AutoModelForCausalLM
235
+ tokenizer_type: AutoTokenizer
236
+
237
+ plugins:
238
+ - axolotl.integrations.liger.LigerPlugin
239
+ liger_rope: true
240
+ liger_rms_norm: true
241
+ liger_swiglu: true
242
+ liger_fused_linear_cross_entropy: true
243
+
244
+
245
+ load_in_8bit: false
246
+ load_in_4bit: false
247
+ strict: false
248
+
249
+ datasets:
250
+ - path: PocketDoc/Dans-Personamaxx-VN
251
+ type: dan-chat-advanced
252
+ - path: NewEden/vanilla-backrooms-claude-sharegpt
253
+ type: dan-chat-advanced
254
+ - path: PocketDoc/Dans-Prosemaxx-Cowriter-3-S
255
+ type: dan-chat-advanced
256
+ - path: NewEden/Misc-Mang-Sharegpt
257
+ type: dan-chat-advanced
258
+ - path: PocketDoc/Dans-Prosemaxx-Adventure
259
+ type: dan-chat-advanced
260
+ - path: PocketDoc/Dans-Failuremaxx-Adventure-3
261
+ type: dan-chat-advanced
262
+ - path: PocketDoc/Dans-Prosemaxx-InstructWriter-ZeroShot
263
+ type: dan-chat-advanced
264
+ - path: PocketDoc/Dans-Prosemaxx-InstructWriter-Continue
265
+ type: dan-chat-advanced
266
+
267
+ shuffle_merged_datasets: true
268
+ dataset_prepared_path: prepared_data
269
+ val_set_size: 0.0
270
+ output_dir: ./phi4-adventure
271
+
272
+ sequence_len: 16384
273
+ sample_packing: true
274
+ pad_to_sequence_len: true
275
+
276
+ adapter: lora
277
+ lora_model_dir:
278
+ lora_r: 128
279
+ lora_alpha: 16
280
+ lora_dropout: 0.05
281
+ lora_target_modules:
282
+ - gate_proj
283
+ - down_proj
284
+ - up_proj
285
+ - q_proj
286
+ - v_proj
287
+ - k_proj
288
+ - o_proj
289
+
290
+ lora_modules_to_save:
291
+ - embed_tokens
292
+ - lm_head
293
+
294
+
295
+ wandb_project: mag-phi
296
+ wandb_entity:
297
+ wandb_watch:
298
+ wandb_name: adventure-attempt-01
299
+ wandb_log_model:
300
+
301
+ gradient_accumulation_steps: 4
302
+ micro_batch_size: 2
303
+ num_epochs: 2
304
+ optimizer: paged_ademamix_8bit
305
+ lr_scheduler: cosine
306
+ learning_rate: 2e-5
307
+
308
+ train_on_inputs: false
309
+ group_by_length: false
310
+ bf16: auto
311
+ fp16:
312
+ tf32: false
313
+
314
+ gradient_checkpointing: unsloth
315
+ early_stopping_patience:
316
+ resume_from_checkpoint:
317
+ local_rank:
318
+ logging_steps: 1
319
+ xformers_attention:
320
+ flash_attention: true
321
+
322
+ warmup_steps: 15
323
+ evals_per_epoch: 4
324
+ eval_table_size:
325
+ eval_max_new_tokens: 128
326
+ saves_per_epoch: 2
327
+ debug:
328
+ deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
329
+ weight_decay: 0.01
330
+ fsdp:
331
+ fsdp_config:
332
  ```
333
+
334
+ </details>
335
+ </div>
336
+
337
+ <div align="center">
338
+
339
+ <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);">
340
+
341
+ ## ⚡ Credits
342
+ <div style="display: flex; justify-content: center;">
343
+ <div style="display: grid; grid-template-columns: repeat(auto-fit, minmax(200px, 1fr)); gap: 10px; margin: 20px 0; max-width: 600px;">
344
+
345
+ <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;">
346
+ <a href="https://huggingface.co/lucyknada">
347
+ <img src="https://img.shields.io/badge/%F0%9F%8C%9F-Lucy_Knada-blueviolet" alt="Lucy Knada">
348
+ </a>
349
+ </div>
350
+
351
+ <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;">
352
+ <a href="https://huggingface.co/jeiku">
353
+ <img src="https://img.shields.io/badge/%E2%9A%94%EF%B8%8F-jeiku-blueviolet" alt="jeiku">
354
+ </a>
355
+ </div>
356
+
357
+ <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;">
358
+ <a href="https://huggingface.co/intervitens">
359
+ <img src="https://img.shields.io/badge/%F0%9F%9B%A1%EF%B8%8F-Intervitens-blueviolet" alt="Intervitens">
360
+ </a>
361
+ </div>
362
+
363
+ <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;">
364
+ <a href="https://huggingface.co/kalomaze">
365
+ <img src="https://img.shields.io/badge/%F0%9F%94%AE-Kalomaze-blueviolet" alt="Kalomaze">
366
+ </a>
367
+ </div>
368
+
369
+ <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;">
370
+ <a href="https://huggingface.co/kubernetes-bad">
371
+ <img src="https://img.shields.io/badge/%E2%9A%A1-Kubernetes_Bad-blueviolet" alt="Kubernetes Bad">
372
+ </a>
373
+ </div>
374
+
375
+ <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;">
376
+ <a href="https://huggingface.co/anthracite-org">
377
+ <img src="https://img.shields.io/badge/%F0%9F%8C%91-Anthracite-blueviolet" alt="Anthracite">
378
+ </a>
379
+ </div>
380
+ </div>
381
+ </div>
382
+ </div>
383
+
384
+ ---
385
+
386
+ <div align="center">
387
+ <div style="font-size:0.8em; opacity:0.8;">Made by</div>
388
+ <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>
389
+ </div>
390
+
391
+ </div>
392
+ </body>
393
+ </html>