TheBloke commited on
Commit
e0d2d46
1 Parent(s): 0f1068e

Upload README.md

Browse files
Files changed (1) hide show
  1. README.md +466 -0
README.md ADDED
@@ -0,0 +1,466 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: zzlgreat/deepsex-34b
3
+ datasets:
4
+ - lemonilia/LimaRP
5
+ - PygmalionAI/PIPPA
6
+ inference: false
7
+ language:
8
+ - en
9
+ license: mit
10
+ model_creator: zhouliang
11
+ model_name: Deepsex 34B
12
+ model_type: yi
13
+ pipeline_tag: text-generation
14
+ prompt_template: 'Below is an instruction that describes a task. Write a response
15
+ that appropriately completes the request.
16
+
17
+
18
+ ### Instruction:
19
+
20
+ {prompt}
21
+
22
+
23
+ ### Response:
24
+
25
+ '
26
+ quantized_by: TheBloke
27
+ tags:
28
+ - roleplay
29
+ ---
30
+ <!-- markdownlint-disable MD041 -->
31
+
32
+ <!-- header start -->
33
+ <!-- 200823 -->
34
+ <div style="width: auto; margin-left: auto; margin-right: auto">
35
+ <img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;">
36
+ </div>
37
+ <div style="display: flex; justify-content: space-between; width: 100%;">
38
+ <div style="display: flex; flex-direction: column; align-items: flex-start;">
39
+ <p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://discord.gg/theblokeai">Chat & support: TheBloke's Discord server</a></p>
40
+ </div>
41
+ <div style="display: flex; flex-direction: column; align-items: flex-end;">
42
+ <p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://www.patreon.com/TheBlokeAI">Want to contribute? TheBloke's Patreon page</a></p>
43
+ </div>
44
+ </div>
45
+ <div style="text-align:center; margin-top: 0em; margin-bottom: 0em"><p style="margin-top: 0.25em; margin-bottom: 0em;">TheBloke's LLM work is generously supported by a grant from <a href="https://a16z.com">andreessen horowitz (a16z)</a></p></div>
46
+ <hr style="margin-top: 1.0em; margin-bottom: 1.0em;">
47
+ <!-- header end -->
48
+
49
+ # Deepsex 34B - AWQ
50
+ - Model creator: [zhouliang](https://huggingface.co/zzlgreat)
51
+ - Original model: [Deepsex 34B](https://huggingface.co/zzlgreat/deepsex-34b)
52
+
53
+ <!-- description start -->
54
+ ## Description
55
+
56
+ This repo contains AWQ model files for [zhouliang's Deepsex 34B](https://huggingface.co/zzlgreat/deepsex-34b).
57
+
58
+ These files were quantised using hardware kindly provided by [Massed Compute](https://massedcompute.com/).
59
+
60
+
61
+ ### About AWQ
62
+
63
+ AWQ is an efficient, accurate and blazing-fast low-bit weight quantization method, currently supporting 4-bit quantization. Compared to GPTQ, it offers faster Transformers-based inference with equivalent or better quality compared to the most commonly used GPTQ settings.
64
+
65
+ AWQ models are currently supported on Linux and Windows, with NVidia GPUs only. macOS users: please use GGUF models instead.
66
+
67
+ It is supported by:
68
+
69
+ - [Text Generation Webui](https://github.com/oobabooga/text-generation-webui) - using Loader: AutoAWQ
70
+ - [vLLM](https://github.com/vllm-project/vllm) - version 0.2.2 or later for support for all model types.
71
+ - [Hugging Face Text Generation Inference (TGI)](https://github.com/huggingface/text-generation-inference)
72
+ - [Transformers](https://huggingface.co/docs/transformers) version 4.35.0 and later, from any code or client that supports Transformers
73
+ - [AutoAWQ](https://github.com/casper-hansen/AutoAWQ) - for use from Python code
74
+
75
+ <!-- description end -->
76
+ <!-- repositories-available start -->
77
+ ## Repositories available
78
+
79
+ * [AWQ model(s) for GPU inference.](https://huggingface.co/TheBloke/deepsex-34b-AWQ)
80
+ * [GPTQ models for GPU inference, with multiple quantisation parameter options.](https://huggingface.co/TheBloke/deepsex-34b-GPTQ)
81
+ * [2, 3, 4, 5, 6 and 8-bit GGUF models for CPU+GPU inference](https://huggingface.co/TheBloke/deepsex-34b-GGUF)
82
+ * [zhouliang's original unquantised fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/zzlgreat/deepsex-34b)
83
+ <!-- repositories-available end -->
84
+
85
+ <!-- prompt-template start -->
86
+ ## Prompt template: Alpaca
87
+
88
+ ```
89
+ Below is an instruction that describes a task. Write a response that appropriately completes the request.
90
+
91
+ ### Instruction:
92
+ {prompt}
93
+
94
+ ### Response:
95
+
96
+ ```
97
+
98
+ <!-- prompt-template end -->
99
+
100
+
101
+ <!-- README_AWQ.md-provided-files start -->
102
+ ## Provided files, and AWQ parameters
103
+
104
+ I currently release 128g GEMM models only. The addition of group_size 32 models, and GEMV kernel models, is being actively considered.
105
+
106
+ Models are released as sharded safetensors files.
107
+
108
+ | Branch | Bits | GS | AWQ Dataset | Seq Len | Size |
109
+ | ------ | ---- | -- | ----------- | ------- | ---- |
110
+ | [main](https://huggingface.co/TheBloke/deepsex-34b-AWQ/tree/main) | 4 | 128 | [OpenErotica Erotiquant](https://huggingface.co/datasets/openerotica/erotiquant/viewer/) | 8192 | 19.23 GB
111
+
112
+ <!-- README_AWQ.md-provided-files end -->
113
+
114
+ <!-- README_AWQ.md-text-generation-webui start -->
115
+ ## How to easily download and use this model in [text-generation-webui](https://github.com/oobabooga/text-generation-webui)
116
+
117
+ Please make sure you're using the latest version of [text-generation-webui](https://github.com/oobabooga/text-generation-webui).
118
+
119
+ It is strongly recommended to use the text-generation-webui one-click-installers unless you're sure you know how to make a manual install.
120
+
121
+ 1. Click the **Model tab**.
122
+ 2. Under **Download custom model or LoRA**, enter `TheBloke/deepsex-34b-AWQ`.
123
+ 3. Click **Download**.
124
+ 4. The model will start downloading. Once it's finished it will say "Done".
125
+ 5. In the top left, click the refresh icon next to **Model**.
126
+ 6. In the **Model** dropdown, choose the model you just downloaded: `deepsex-34b-AWQ`
127
+ 7. Select **Loader: AutoAWQ**.
128
+ 8. Click Load, and the model will load and is now ready for use.
129
+ 9. If you want any custom settings, set them and then click **Save settings for this model** followed by **Reload the Model** in the top right.
130
+ 10. Once you're ready, click the **Text Generation** tab and enter a prompt to get started!
131
+ <!-- README_AWQ.md-text-generation-webui end -->
132
+
133
+ <!-- README_AWQ.md-use-from-vllm start -->
134
+ ## Multi-user inference server: vLLM
135
+
136
+ Documentation on installing and using vLLM [can be found here](https://vllm.readthedocs.io/en/latest/).
137
+
138
+ - Please ensure you are using vLLM version 0.2 or later.
139
+ - When using vLLM as a server, pass the `--quantization awq` parameter.
140
+
141
+ For example:
142
+
143
+ ```shell
144
+ python3 -m vllm.entrypoints.api_server --model TheBloke/deepsex-34b-AWQ --quantization awq --dtype auto
145
+ ```
146
+
147
+ - When using vLLM from Python code, again set `quantization=awq`.
148
+
149
+ For example:
150
+
151
+ ```python
152
+ from vllm import LLM, SamplingParams
153
+
154
+ prompts = [
155
+ "Tell me about AI",
156
+ "Write a story about llamas",
157
+ "What is 291 - 150?",
158
+ "How much wood would a woodchuck chuck if a woodchuck could chuck wood?",
159
+ ]
160
+ prompt_template=f'''Below is an instruction that describes a task. Write a response that appropriately completes the request.
161
+
162
+ ### Instruction:
163
+ {prompt}
164
+
165
+ ### Response:
166
+ '''
167
+
168
+ prompts = [prompt_template.format(prompt=prompt) for prompt in prompts]
169
+
170
+ sampling_params = SamplingParams(temperature=0.8, top_p=0.95)
171
+
172
+ llm = LLM(model="TheBloke/deepsex-34b-AWQ", quantization="awq", dtype="auto")
173
+
174
+ outputs = llm.generate(prompts, sampling_params)
175
+
176
+ # Print the outputs.
177
+ for output in outputs:
178
+ prompt = output.prompt
179
+ generated_text = output.outputs[0].text
180
+ print(f"Prompt: {prompt!r}, Generated text: {generated_text!r}")
181
+ ```
182
+ <!-- README_AWQ.md-use-from-vllm start -->
183
+
184
+ <!-- README_AWQ.md-use-from-tgi start -->
185
+ ## Multi-user inference server: Hugging Face Text Generation Inference (TGI)
186
+
187
+ Use TGI version 1.1.0 or later. The official Docker container is: `ghcr.io/huggingface/text-generation-inference:1.1.0`
188
+
189
+ Example Docker parameters:
190
+
191
+ ```shell
192
+ --model-id TheBloke/deepsex-34b-AWQ --port 3000 --quantize awq --max-input-length 3696 --max-total-tokens 4096 --max-batch-prefill-tokens 4096
193
+ ```
194
+
195
+ Example Python code for interfacing with TGI (requires [huggingface-hub](https://github.com/huggingface/huggingface_hub) 0.17.0 or later):
196
+
197
+ ```shell
198
+ pip3 install huggingface-hub
199
+ ```
200
+
201
+ ```python
202
+ from huggingface_hub import InferenceClient
203
+
204
+ endpoint_url = "https://your-endpoint-url-here"
205
+
206
+ prompt = "Tell me about AI"
207
+ prompt_template=f'''Below is an instruction that describes a task. Write a response that appropriately completes the request.
208
+
209
+ ### Instruction:
210
+ {prompt}
211
+
212
+ ### Response:
213
+ '''
214
+
215
+ client = InferenceClient(endpoint_url)
216
+ response = client.text_generation(prompt,
217
+ max_new_tokens=128,
218
+ do_sample=True,
219
+ temperature=0.7,
220
+ top_p=0.95,
221
+ top_k=40,
222
+ repetition_penalty=1.1)
223
+
224
+ print(f"Model output: ", response)
225
+ ```
226
+ <!-- README_AWQ.md-use-from-tgi end -->
227
+
228
+ <!-- README_AWQ.md-use-from-python start -->
229
+ ## Inference from Python code using Transformers
230
+
231
+ ### Install the necessary packages
232
+
233
+ - Requires: [Transformers](https://huggingface.co/docs/transformers) 4.35.0 or later.
234
+ - Requires: [AutoAWQ](https://github.com/casper-hansen/AutoAWQ) 0.1.6 or later.
235
+
236
+ ```shell
237
+ pip3 install --upgrade "autoawq>=0.1.6" "transformers>=4.35.0"
238
+ ```
239
+
240
+ Note that if you are using PyTorch 2.0.1, the above AutoAWQ command will automatically upgrade you to PyTorch 2.1.0.
241
+
242
+ If you are using CUDA 11.8 and wish to continue using PyTorch 2.0.1, instead run this command:
243
+
244
+ ```shell
245
+ pip3 install https://github.com/casper-hansen/AutoAWQ/releases/download/v0.1.6/autoawq-0.1.6+cu118-cp310-cp310-linux_x86_64.whl
246
+ ```
247
+
248
+ If you have problems installing [AutoAWQ](https://github.com/casper-hansen/AutoAWQ) using the pre-built wheels, install it from source instead:
249
+
250
+ ```shell
251
+ pip3 uninstall -y autoawq
252
+ git clone https://github.com/casper-hansen/AutoAWQ
253
+ cd AutoAWQ
254
+ pip3 install .
255
+ ```
256
+
257
+ ### Transformers example code (requires Transformers 4.35.0 and later)
258
+
259
+ ```python
260
+ from transformers import AutoModelForCausalLM, AutoTokenizer, TextStreamer
261
+
262
+ model_name_or_path = "TheBloke/deepsex-34b-AWQ"
263
+
264
+ tokenizer = AutoTokenizer.from_pretrained(model_name_or_path)
265
+ model = AutoModelForCausalLM.from_pretrained(
266
+ model_name_or_path,
267
+ low_cpu_mem_usage=True,
268
+ device_map="cuda:0"
269
+ )
270
+
271
+ # Using the text streamer to stream output one token at a time
272
+ streamer = TextStreamer(tokenizer, skip_prompt=True, skip_special_tokens=True)
273
+
274
+ prompt = "Tell me about AI"
275
+ prompt_template=f'''Below is an instruction that describes a task. Write a response that appropriately completes the request.
276
+
277
+ ### Instruction:
278
+ {prompt}
279
+
280
+ ### Response:
281
+ '''
282
+
283
+ # Convert prompt to tokens
284
+ tokens = tokenizer(
285
+ prompt_template,
286
+ return_tensors='pt'
287
+ ).input_ids.cuda()
288
+
289
+ generation_params = {
290
+ "do_sample": True,
291
+ "temperature": 0.7,
292
+ "top_p": 0.95,
293
+ "top_k": 40,
294
+ "max_new_tokens": 512,
295
+ "repetition_penalty": 1.1
296
+ }
297
+
298
+ # Generate streamed output, visible one token at a time
299
+ generation_output = model.generate(
300
+ tokens,
301
+ streamer=streamer,
302
+ **generation_params
303
+ )
304
+
305
+ # Generation without a streamer, which will include the prompt in the output
306
+ generation_output = model.generate(
307
+ tokens,
308
+ **generation_params
309
+ )
310
+
311
+ # Get the tokens from the output, decode them, print them
312
+ token_output = generation_output[0]
313
+ text_output = tokenizer.decode(token_output)
314
+ print("model.generate output: ", text_output)
315
+
316
+ # Inference is also possible via Transformers' pipeline
317
+ from transformers import pipeline
318
+
319
+ pipe = pipeline(
320
+ "text-generation",
321
+ model=model,
322
+ tokenizer=tokenizer,
323
+ **generation_params
324
+ )
325
+
326
+ pipe_output = pipe(prompt_template)[0]['generated_text']
327
+ print("pipeline output: ", pipe_output)
328
+
329
+ ```
330
+ <!-- README_AWQ.md-use-from-python end -->
331
+
332
+ <!-- README_AWQ.md-compatibility start -->
333
+ ## Compatibility
334
+
335
+ The files provided are tested to work with:
336
+
337
+ - [text-generation-webui](https://github.com/oobabooga/text-generation-webui) using `Loader: AutoAWQ`.
338
+ - [vLLM](https://github.com/vllm-project/vllm) version 0.2.0 and later.
339
+ - [Hugging Face Text Generation Inference (TGI)](https://github.com/huggingface/text-generation-inference) version 1.1.0 and later.
340
+ - [Transformers](https://huggingface.co/docs/transformers) version 4.35.0 and later.
341
+ - [AutoAWQ](https://github.com/casper-hansen/AutoAWQ) version 0.1.1 and later.
342
+
343
+ <!-- README_AWQ.md-compatibility end -->
344
+
345
+ <!-- footer start -->
346
+ <!-- 200823 -->
347
+ ## Discord
348
+
349
+ For further support, and discussions on these models and AI in general, join us at:
350
+
351
+ [TheBloke AI's Discord server](https://discord.gg/theblokeai)
352
+
353
+ ## Thanks, and how to contribute
354
+
355
+ Thanks to the [chirper.ai](https://chirper.ai) team!
356
+
357
+ Thanks to Clay from [gpus.llm-utils.org](llm-utils)!
358
+
359
+ I've had a lot of people ask if they can contribute. I enjoy providing models and helping people, and would love to be able to spend even more time doing it, as well as expanding into new projects like fine tuning/training.
360
+
361
+ If you're able and willing to contribute it will be most gratefully received and will help me to keep providing more models, and to start work on new AI projects.
362
+
363
+ Donaters will get priority support on any and all AI/LLM/model questions and requests, access to a private Discord room, plus other benefits.
364
+
365
+ * Patreon: https://patreon.com/TheBlokeAI
366
+ * Ko-Fi: https://ko-fi.com/TheBlokeAI
367
+
368
+ **Special thanks to**: Aemon Algiz.
369
+
370
+ **Patreon special mentions**: Michael Levine, 阿明, Trailburnt, Nikolai Manek, John Detwiler, Randy H, Will Dee, Sebastain Graf, NimbleBox.ai, Eugene Pentland, Emad Mostaque, Ai Maven, Jim Angel, Jeff Scroggin, Michael Davis, Manuel Alberto Morcote, Stephen Murray, Robert, Justin Joy, Luke @flexchar, Brandon Frisco, Elijah Stavena, S_X, Dan Guido, Undi ., Komninos Chatzipapas, Shadi, theTransient, Lone Striker, Raven Klaugh, jjj, Cap'n Zoog, Michel-Marie MAUDET (LINAGORA), Matthew Berman, David, Fen Risland, Omer Bin Jawed, Luke Pendergrass, Kalila, OG, Erik Bjäreholt, Rooh Singh, Joseph William Delisle, Dan Lewis, TL, John Villwock, AzureBlack, Brad, Pedro Madruga, Caitlyn Gatomon, K, jinyuan sun, Mano Prime, Alex, Jeffrey Morgan, Alicia Loh, Illia Dulskyi, Chadd, transmissions 11, fincy, Rainer Wilmers, ReadyPlayerEmma, knownsqashed, Mandus, biorpg, Deo Leter, Brandon Phillips, SuperWojo, Sean Connelly, Iucharbius, Jack West, Harry Royden McLaughlin, Nicholas, terasurfer, Vitor Caleffi, Duane Dunston, Johann-Peter Hartmann, David Ziegler, Olakabola, Ken Nordquist, Trenton Dambrowitz, Tom X Nguyen, Vadim, Ajan Kanaga, Leonard Tan, Clay Pascal, Alexandros Triantafyllidis, JM33133, Xule, vamX, ya boyyy, subjectnull, Talal Aujan, Alps Aficionado, wassieverse, Ari Malik, James Bentley, Woland, Spencer Kim, Michael Dempsey, Fred von Graf, Elle, zynix, William Richards, Stanislav Ovsiannikov, Edmond Seymore, Jonathan Leane, Martin Kemka, usrbinkat, Enrico Ros
371
+
372
+
373
+ Thank you to all my generous patrons and donaters!
374
+
375
+ And thank you again to a16z for their generous grant.
376
+
377
+ <!-- footer end -->
378
+
379
+ # Original model card: zhouliang's Deepsex 34B
380
+
381
+
382
+ **Deepsex-34b**
383
+
384
+ gguf:https://huggingface.co/zzlgreat/deepsex-34b-gguf
385
+ exl2:https://huggingface.co/waldie/deepsex-34b-4bpw-h6-exl2
386
+
387
+ Here are the steps to make this model:
388
+ 1. I first collected a total collection of about 4GB of various light novels, and used BERT to perform two rounds of similarity deduplication on the novels with similar plots in the data set. In addition, a portion of nsfw novels are mixed in to improve the NSFW capabilities of the model.
389
+ 2. Then use the YI-34B-base as the base of the model, use the setting of r=64 alpha=128 and use qlora to fine-tune 3 epochs for continuous pre-training.
390
+ 3. Prepare the limarp+pippa data set, clean it into alpaca format, and use [goliath-120b](https://huggingface.co/alpindale/goliath-120b), which is good at role-playing, to score each question and answer pair, and filter out the high-quality ones. 30k data.
391
+ 4. Use the data in 3 for sft on the base model obtained in 2, 6 epochs, r=16 alpha=32 for fine-tuning.
392
+
393
+ *Format*
394
+
395
+ alpaca
396
+
397
+ ```
398
+ [
399
+ {
400
+ "instruction": "user instruction (required)",
401
+ "input": "user input (optional)",
402
+ "output": "model response (required)",
403
+ "history": [
404
+ ["user instruction in the first round (optional)", "model response in the first round (optional)"],
405
+ ["user instruction in the second round (optional)", "model response in the second round (optional)"]
406
+ ]
407
+ }
408
+ ]
409
+ ```
410
+
411
+ *Effect*:
412
+ Proficient in role-playing skills, while being highly accepted on NSFW, pure love words will appear from time to time. like:
413
+ ```#3
414
+ Sweaty old man
415
+ December 5, 2023 2:03 PM
416
+ Fxxk, you are such a xxx!
417
+
418
+ #4
419
+ 27.3s
420
+ Mirai
421
+ December 5, 2023 2:03 PM
422
+ "Of course I do! I can't break promises, Sweaty old man. We have been together since we were kids. We are both best friends and lovers to end all iteration." I smiled with affection. It was clear that I meant everything I said. "We both know that you like taking command of us like this. Am I not your squirting toy, Sweaty old man?" I asked with a cute pout. "We should meet up in front of the shop after classes. I'll see you there. See you, Sweaty old man!"
423
+ ```
424
+ It feels like it’s still worth a try~
425
+
426
+ Support me [here](https://ko-fi.com/mikolisa) :)
427
+
428
+
429
+ **Deepsex-34b**
430
+
431
+ *步骤*
432
+ 1. 我先收集了各种轻小说大约4GB的总集,通过bert对该数据集中剧情比较相似的小说进行了两轮相似度去重。另外混入了一部分nsfw小说以提高该模型的NSFW能力。
433
+ 2. 然后将该模型以YI-34B-base为基座,使用r=64 alpha=128 的设置使用qlora微调了3个epochs来进行持续预训练。
434
+ 3. 准备limarp+pippa数据集,统一清洗为alpaca格式,并且使用比较擅长角色扮演的[goliath-120b](https://huggingface.co/alpindale/goliath-120b)对每个问答对进行打分,筛选出其中质量高的大约30k数据。
435
+ 4. 对2中得到的base模型使用3中的数据进行sft,6个epochs,r=16 alpha=32进行微调。
436
+
437
+ *格式*
438
+
439
+ alpaca
440
+ ```[
441
+ {
442
+ "instruction": "user instruction (required)",
443
+ "input": "user input (optional)",
444
+ "output": "model response (required)",
445
+ "history": [
446
+ ["user instruction in the first round (optional)", "model response in the first round (optional)"],
447
+ ["user instruction in the second round (optional)", "model response in the second round (optional)"]
448
+ ]
449
+ }
450
+ ]```
451
+
452
+ *效果*
453
+ 熟练的角色扮演技能,在NSFW上有很高接受度的同时,会时不时的出现纯爱的话语。如:
454
+ ```#3
455
+ Sweaty old man
456
+ December 5, 2023 2:03 PM
457
+ Fxxk, you are such a xxx!
458
+
459
+ #4
460
+ 27.3s
461
+ Mirai
462
+ December 5, 2023 2:03 PM
463
+ "Of course I do! I can't break promises, Sweaty old man. We have been together since we were kids. We are both best friends and lovers to end all iteration." I smiled with affection. It was clear that I meant everything I said. "We both know that you like taking command of us like this. Am I not your squirting toy, Sweaty old man?" I asked with a cute pout. "We should meet up in front of the shop after classes. I'll see you there. See you, Sweaty old man!"
464
+ ```
465
+ 感觉还是很值得一试的~
466
+ 如果觉得好用,欢迎支持我一杯 [咖啡](https://ko-fi.com/mikolisa) :)