kernalkue commited on
Commit
0692817
·
verified ·
1 Parent(s): 3181ed9

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +3 -209
README.md CHANGED
@@ -1,219 +1,13 @@
1
  ---
2
- base_model: [BruhzWater/Sapphira-L3.3-70b-0.1]
3
- library_name: transformers
4
  base_model_relation: quantized
5
  tags:
6
  - exl3
7
  - exllamav3
8
 
9
  ---
10
- # Sapphira-L3.3-70b-0.1
11
 
12
- ![image/png](https://cdn-uploads.huggingface.co/production/uploads/66ca56e62400073af3ad2972/CPUXeq81a9o0_ClXCcG68.png)
13
 
14
- Storytelling and RP model with increased coherence, thanks to cogito-v2-preview-llama-70B.
15
 
16
- iMatrix quants: https://huggingface.co/mradermacher/Sapphira-L3.3-70b-0.1-i1-GGUF
17
-
18
- Static quants: https://huggingface.co/mradermacher/Sapphira-L3.3-70b-0.1-GGUF
19
-
20
- Chat Template:
21
- -
22
- Llama3
23
-
24
- Instruction Template:
25
- -
26
- Deep Cogito
27
-
28
- Llama3
29
-
30
- Sampler Settings
31
- -
32
-
33
- Starter:
34
- ```
35
- Temp: 1
36
- Min_P: 0.02
37
- Top_P: 1
38
- ```
39
-
40
- Experimental 1:
41
- ```
42
- Temp: .95 - 1.1
43
- Min_P: .015 - .03
44
- Top_P: .97 - .99
45
- XTC_Threshold: .11
46
- XTC_Probability: .15
47
- ```
48
-
49
- Experimental 2:
50
- ```
51
- Temp: .95 - 1.1
52
- Min_P: .015 - .03
53
- Top_P: 1
54
- Typical_P: .99
55
- XTC_Threshold: .11
56
- XTC_Probability: .15
57
- ```
58
-
59
- ### Merge Method
60
-
61
- This model was merged using the [Multi-SLERP](https://goddard.blog/posts/multislerp-wow-what-a-cool-idea) merge method using deepcogito--cogito-v2-preview-llama-70B as a base.
62
-
63
- ### Models Merged
64
-
65
- The following models were included in the merge:
66
- * BruhzWater--Apocrypha-L3.3-70b-0.3
67
- * BruhzWater--Serpents-Tongue-L3.3-70b-0.3
68
-
69
- ### Configuration
70
-
71
- The following YAML configuration was used to produce this model:
72
-
73
- ```yaml
74
- models:
75
- - model: /workspace/cache/models--BruhzWater--Apocrypha-L3.3-70b-0.3/snapshots/3facb4c0a7b953ff34a5caa90976830bf82a84c2
76
- parameters:
77
- weight: [0.5]
78
- - model: /workspace/cache/models--BruhzWater--Serpents-Tongue-L3.3-70b-0.3/snapshots/d007a7bcc7047d712abb2dfb6ad940fe03cd2047
79
- parameters:
80
- weight: [0.5]
81
- base_model: /workspace/cache/models--deepcogito--cogito-v2-preview-llama-70B/snapshots/1e1d12e8eaebd6084a8dcf45ecdeaa2f4b8879ce
82
- merge_method: multislerp
83
- tokenizer:
84
- source: base
85
- chat_template: llama3
86
- parameters:
87
- normalize_weights: false
88
- eps: 1e-9
89
- pad_to_multiple_of: 8
90
- int8_mask: true
91
- dtype: bfloat16
92
- ```
93
-
94
- ### Instruct Template
95
-
96
- Deep Cogito
97
-
98
- ```
99
- {{- '<|begin_of_text|>' }}
100
- {%- if not tools is defined %}
101
- {%- set tools = none %}
102
- {%- endif %}
103
- {%- if not enable_thinking is defined %}
104
- {%- set enable_thinking = false %}
105
- {%- endif %}
106
- {#- This block extracts the system message, so we can slot it into the right place. #}
107
- {%- if messages[0]['role'] == 'system' %}
108
- {%- set system_message = messages[0]['content']|trim %}
109
- {%- set messages = messages[1:] %}
110
- {%- else %}
111
- {%- set system_message = "" %}
112
- {%- endif %}
113
- {#- Set the system message. If enable_thinking is true, add the "Enable deep thinking subroutine." #}
114
- {%- if enable_thinking %}
115
- {%- if system_message != "" %}
116
- {%- set system_message = "Enable deep thinking subroutine.
117
-
118
- " ~ system_message %}
119
- {%- else %}
120
- {%- set system_message = "Enable deep thinking subroutine." %}
121
- {%- endif %}
122
- {%- endif %}
123
- {#- Set the system message. In case there are tools present, add them to the system message. #}
124
- {%- if tools is not none or system_message != '' %}
125
- {{- "<|start_header_id|>system<|end_header_id|>
126
-
127
- " }}
128
- {{- system_message }}
129
- {%- if tools is not none %}
130
- {%- if system_message != "" %}
131
- {{- "
132
-
133
- " }}
134
- {%- endif %}
135
- {{- "Available Tools:
136
- " }}
137
- {%- for t in tools %}
138
- {{- t | tojson(indent=4) }}
139
- {{- "
140
-
141
- " }}
142
- {%- endfor %}
143
- {%- endif %}
144
- {{- "<|eot_id|>" }}
145
- {%- endif %}
146
-
147
- {#- Rest of the messages #}
148
- {%- for message in messages %}
149
- {#- The special cases are when the message is from a tool (via role ipython/tool/tool_results) or when the message is from the assistant, but has "tool_calls". If not, we add the message directly as usual. #}
150
- {#- Case 1 - Usual, non tool related message. #}
151
- {%- if not (message.role == "ipython" or message.role == "tool" or message.role == "tool_results" or (message.tool_calls is defined and message.tool_calls is not none)) %}
152
- {{- '<|start_header_id|>' + message['role'] + '<|end_header_id|>
153
-
154
- ' }}
155
- {%- if message['content'] is string %}
156
- {{- message['content'] | trim }}
157
- {%- else %}
158
- {%- for item in message['content'] %}
159
- {%- if item.type == 'text' %}
160
- {{- item.text | trim }}
161
- {%- endif %}
162
- {%- endfor %}
163
- {%- endif %}
164
- {{- '<|eot_id|>' }}
165
-
166
- {#- Case 2 - the response is from the assistant, but has a tool call returned. The assistant may also have returned some content along with the tool call. #}
167
- {%- elif message.tool_calls is defined and message.tool_calls is not none %}
168
- {{- "<|start_header_id|>assistant<|end_header_id|>
169
-
170
- " }}
171
- {%- if message['content'] is string %}
172
- {{- message['content'] | trim }}
173
- {%- else %}
174
- {%- for item in message['content'] %}
175
- {%- if item.type == 'text' %}
176
- {{- item.text | trim }}
177
- {%- if item.text | trim != "" %}
178
- {{- "
179
-
180
- " }}
181
- {%- endif %}
182
- {%- endif %}
183
- {%- endfor %}
184
- {%- endif %}
185
- {{- "[" }}
186
- {%- for tool_call in message.tool_calls %}
187
- {%- set out = tool_call.function|tojson %}
188
- {%- if not tool_call.id is defined %}
189
- {{- out }}
190
- {%- else %}
191
- {{- out[:-1] }}
192
- {{- ', "id": "' + tool_call.id + '"}' }}
193
- {%- endif %}
194
- {%- if not loop.last %}
195
- {{- ", " }}
196
- {%- else %}
197
- {{- "]<|eot_id|>" }}
198
- {%- endif %}
199
- {%- endfor %}
200
-
201
- {#- Case 3 - the response is from a tool call. The tool call may have an id associated with it as well. If it does, we add it to the prompt. #}
202
- {%- elif message.role == "ipython" or message["role"] == "tool_results" or message["role"] == "tool" %}
203
- {{- "<|start_header_id|>ipython<|end_header_id|>
204
-
205
- " }}
206
- {%- if message.tool_call_id is defined and message.tool_call_id != '' %}
207
- {{- '{"content": ' + (message.content | tojson) + ', "call_id": "' + message.tool_call_id + '"}' }}
208
- {%- else %}
209
- {{- '{"content": ' + (message.content | tojson) + '}' }}
210
- {%- endif %}
211
- {{- "<|eot_id|>" }}
212
- {%- endif %}
213
- {%- endfor %}
214
- {%- if add_generation_prompt %}
215
- {{- '<|start_header_id|>assistant<|end_header_id|>
216
-
217
- ' }}
218
- {%- endif %}
219
- ```
 
1
  ---
2
+ base_model: BruhzWater/Sapphira-L3.3-70b-0.1
 
3
  base_model_relation: quantized
4
  tags:
5
  - exl3
6
  - exllamav3
7
 
8
  ---
 
9
 
10
+ ## 6.0bpw EXL3 Quant of BruhzWater/Sapphira-L3.3-70b-0.1
11
 
12
+ 6.0bpw H6 EXL3 quant of [BruhzWater/Sapphira-L3.3-70b-0.1](https://huggingface.co/BruhzWater/Sapphira-L3.3-70b-0.1) using <a href="https://github.com/turboderp-org/exllamav3/">exllamav3</a> for quantization.
13