Casual-Autopsy
commited on
Commit
•
7410758
1
Parent(s):
8840285
Update README.md
Browse files
README.md
CHANGED
@@ -11,11 +11,8 @@ tags:
|
|
11 |
- rp
|
12 |
- roleplay
|
13 |
- role-play
|
14 |
-
- chain-of-thoughts
|
15 |
- summarization
|
16 |
- emotion classification
|
17 |
-
- biology
|
18 |
-
- psychology
|
19 |
base_model:
|
20 |
- nothingiisreal/L3-8B-Celeste-v1
|
21 |
- Nitral-AI/Hathor_Tahsin-L3-8B-v0.85
|
@@ -33,6 +30,14 @@ base_model:
|
|
33 |
- OEvortex/Emotional-llama-8B
|
34 |
- lighteternal/Llama3-merge-biomed-8b
|
35 |
- Casual-Autopsy/Llama3-merge-psychotherapy-8b
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
36 |
---
|
37 |
| <img src="https://huggingface.co/Casual-Autopsy/L3-Super-Nova-RP-8B/resolve/main/Card-Assets/NovaKid-Girl.jpeg" width="50%" height="50%" style="display: block; margin: auto;"> |
|
38 |
|:---:|
|
@@ -45,10 +50,14 @@ base_model:
|
|
45 |
***
|
46 |
***
|
47 |
## Presets
|
48 |
-
|
|
|
|
|
|
|
|
|
49 |
```yaml
|
50 |
-
Top K:
|
51 |
-
|
52 |
Repetition Penalty: 1.01
|
53 |
# Don't make this higher, DRY handles the bulk of Squashing Repetition.
|
54 |
# This is justs to lightly nudge the bot to move the plot forward
|
@@ -68,6 +77,10 @@ Dynamic Temperature:
|
|
68 |
Exponent: 0.85
|
69 |
```
|
70 |
|
|
|
|
|
|
|
|
|
71 |
***
|
72 |
***
|
73 |
## Usage Info
|
@@ -110,6 +123,14 @@ The following models were used to make this merge:
|
|
110 |
* [OEvortex/Emotional-llama-8B](https://huggingface.co/OEvortex/Emotional-llama-8B)
|
111 |
* [lighteternal/Llama3-merge-biomed-8b](https://huggingface.co/lighteternal/Llama3-merge-biomed-8b)
|
112 |
* [Casual-Autopsy/Llama3-merge-psychotherapy-8b](https://huggingface.co/Casual-Autopsy/Llama3-merge-psychotherapy-8b)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
113 |
|
114 |
***
|
115 |
***
|
@@ -118,8 +139,6 @@ The following models were used to make this merge:
|
|
118 |
***
|
119 |
### [Open LLM Leaderboard](https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard)
|
120 |
|
121 |
-
Detailed results can be found [here](https://huggingface.co/datasets/open-llm-leaderboard/details_Casual-Autopsy__L3-Umbral-Mind-RP-v2.0-8B)
|
122 |
-
|
123 |
**Explaination for AI RP newbies:** IFEval is the most important evaluation for RP AIs as it determines how well it can follow OOC, Lorebooks, and most importantly character cards.
|
124 |
The rest don't matter. At least not nearly as much as IFEval.
|
125 |
|
@@ -158,42 +177,167 @@ The following YAML configs were used to make this merge.
|
|
158 |
### Super-Nova-CRE_pt.1
|
159 |
|
160 |
```yaml
|
161 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
162 |
```
|
163 |
|
164 |
***
|
165 |
### Super-Nova-CRE_pt.2
|
166 |
|
167 |
```yaml
|
168 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
169 |
```
|
170 |
|
171 |
***
|
172 |
### Super-Nova-UNC_pt.1
|
173 |
|
174 |
```yaml
|
175 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
176 |
```
|
177 |
|
178 |
***
|
179 |
### Super-Nova-UNC_pt.2
|
180 |
|
181 |
```yaml
|
182 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
183 |
```
|
184 |
|
185 |
***
|
186 |
### Super-Nova-INT_pt.1
|
187 |
|
188 |
```yaml
|
189 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
190 |
```
|
191 |
|
192 |
***
|
193 |
### Super-Nova-INT_pt.2
|
194 |
|
195 |
```yaml
|
196 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
197 |
```
|
198 |
|
199 |
***
|
@@ -201,39 +345,183 @@ The following YAML configs were used to make this merge.
|
|
201 |
|
202 |
```yaml
|
203 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
204 |
```
|
205 |
|
206 |
***
|
207 |
### Super-Nova-UNC
|
208 |
|
209 |
```yaml
|
210 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
211 |
```
|
212 |
|
213 |
***
|
214 |
### Super-Nova-INT
|
215 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
216 |
```yaml
|
217 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
218 |
```
|
219 |
|
220 |
***
|
221 |
-
### Super-Nova-
|
222 |
|
223 |
```yaml
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
224 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
225 |
```
|
226 |
|
227 |
***
|
228 |
### Super-Nova-RP_pt.2
|
229 |
|
230 |
```yaml
|
231 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
232 |
```
|
233 |
|
234 |
***
|
235 |
### L3-Super-Nova-RP-8B
|
236 |
|
237 |
```yaml
|
238 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
239 |
```
|
|
|
11 |
- rp
|
12 |
- roleplay
|
13 |
- role-play
|
|
|
14 |
- summarization
|
15 |
- emotion classification
|
|
|
|
|
16 |
base_model:
|
17 |
- nothingiisreal/L3-8B-Celeste-v1
|
18 |
- Nitral-AI/Hathor_Tahsin-L3-8B-v0.85
|
|
|
30 |
- OEvortex/Emotional-llama-8B
|
31 |
- lighteternal/Llama3-merge-biomed-8b
|
32 |
- Casual-Autopsy/Llama3-merge-psychotherapy-8b
|
33 |
+
- Sao10K/L3-8B-Tamamo-v1
|
34 |
+
- ResplendentAI/Nymph_8B
|
35 |
+
- ChaoticNeutrals/T-900-8B
|
36 |
+
- Sao10K/L3-8B-Niitama-v1
|
37 |
+
- bluuwhale/L3-SthenoMaidBlackroot-8B-V1
|
38 |
+
- Hastagaras/Jamet-8B-L3-MK.V-Blackroot
|
39 |
+
- Hastagaras/Halu-8B-Llama3-Blackroot
|
40 |
+
- crestf411/L3-8B-sunfall-v0.4-stheno-v3.2
|
41 |
---
|
42 |
| <img src="https://huggingface.co/Casual-Autopsy/L3-Super-Nova-RP-8B/resolve/main/Card-Assets/NovaKid-Girl.jpeg" width="50%" height="50%" style="display: block; margin: auto;"> |
|
43 |
|:---:|
|
|
|
50 |
***
|
51 |
***
|
52 |
## Presets
|
53 |
+
|
54 |
+
***
|
55 |
+
### Text Gen
|
56 |
+
The Current good staring preset for this model. **Subject to change.**
|
57 |
+
**Settings by yours truly**
|
58 |
```yaml
|
59 |
+
Top K: 40
|
60 |
+
Min P: 0.075
|
61 |
Repetition Penalty: 1.01
|
62 |
# Don't make this higher, DRY handles the bulk of Squashing Repetition.
|
63 |
# This is justs to lightly nudge the bot to move the plot forward
|
|
|
77 |
Exponent: 0.85
|
78 |
```
|
79 |
|
80 |
+
***
|
81 |
+
### Context/Instruct
|
82 |
+
[Virt-io's SillyTavern](https://huggingface.co/Virt-io/SillyTavern-Presets) Presets work really well with this.
|
83 |
+
|
84 |
***
|
85 |
***
|
86 |
## Usage Info
|
|
|
123 |
* [OEvortex/Emotional-llama-8B](https://huggingface.co/OEvortex/Emotional-llama-8B)
|
124 |
* [lighteternal/Llama3-merge-biomed-8b](https://huggingface.co/lighteternal/Llama3-merge-biomed-8b)
|
125 |
* [Casual-Autopsy/Llama3-merge-psychotherapy-8b](https://huggingface.co/Casual-Autopsy/Llama3-merge-psychotherapy-8b)
|
126 |
+
* [Sao10K/L3-8B-Tamamo-v1](https://huggingface.co/Sao10K/L3-8B-Tamamo-v1)
|
127 |
+
* [ResplendentAI/Nymph_8B](https://huggingface.co/ResplendentAI/Nymph_8B)
|
128 |
+
* [ChaoticNeutrals/T-900-8B](https://huggingface.co/ChaoticNeutrals/T-900-8B)
|
129 |
+
* [Sao10K/L3-8B-Niitama-v1](https://huggingface.co/Sao10K/L3-8B-Niitama-v1)
|
130 |
+
* [bluuwhale/L3-SthenoMaidBlackroot-8B-V1](https://huggingface.co/bluuwhale/L3-SthenoMaidBlackroot-8B-V1)
|
131 |
+
* [Hastagaras/Jamet-8B-L3-MK.V-Blackroot](https://huggingface.co/Hastagaras/Jamet-8B-L3-MK.V-Blackroot)
|
132 |
+
* [Hastagaras/Halu-8B-Llama3-Blackroot](https://huggingface.co/Hastagaras/Halu-8B-Llama3-Blackroot)
|
133 |
+
* [crestf411/L3-8B-sunfall-v0.4-stheno-v3.2](https://huggingface.co/crestf411/L3-8B-sunfall-v0.4-stheno-v3.2)
|
134 |
|
135 |
***
|
136 |
***
|
|
|
139 |
***
|
140 |
### [Open LLM Leaderboard](https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard)
|
141 |
|
|
|
|
|
142 |
**Explaination for AI RP newbies:** IFEval is the most important evaluation for RP AIs as it determines how well it can follow OOC, Lorebooks, and most importantly character cards.
|
143 |
The rest don't matter. At least not nearly as much as IFEval.
|
144 |
|
|
|
177 |
### Super-Nova-CRE_pt.1
|
178 |
|
179 |
```yaml
|
180 |
+
models:
|
181 |
+
- model: nothingiisreal/L3-8B-Celeste-v1
|
182 |
+
- model: Nitral-AI/Hathor_Tahsin-L3-8B-v0.85
|
183 |
+
parameters:
|
184 |
+
density: [0.35, 0.45, 0.5, 0.55, 0.65, 0.55, 0.5, 0.45, 0.35]
|
185 |
+
weight: [0.495, 0.165, 0.165, 0.495, 0.495, 0.165, 0.165, 0.495]
|
186 |
+
- model: Sao10K/L3-8B-Stheno-v3.2
|
187 |
+
parameters:
|
188 |
+
density: [0.65, 0.55, 0.5, 0.45, 0.35, 0.45, 0.5, 0.55, 0.65]
|
189 |
+
weight: [0.165, 0.495, 0.495, 0.165, 0.165, 0.495, 0.495, 0.165]
|
190 |
+
merge_method: dare_ties
|
191 |
+
base_model: nothingiisreal/L3-8B-Celeste-v1
|
192 |
+
parameters:
|
193 |
+
normalize: false
|
194 |
+
int8_mask: true
|
195 |
+
dtype: float32
|
196 |
+
out_dtype: bfloat16
|
197 |
```
|
198 |
|
199 |
***
|
200 |
### Super-Nova-CRE_pt.2
|
201 |
|
202 |
```yaml
|
203 |
+
models:
|
204 |
+
- model: nothingiisreal/L3-8B-Celeste-v1
|
205 |
+
- model: ChaoticNeutrals/Poppy_Porpoise-1.0-L3-8B
|
206 |
+
parameters:
|
207 |
+
density: [0.35, 0.45, 0.5, 0.55, 0.65, 0.55, 0.5, 0.45, 0.35]
|
208 |
+
weight: [0.165, 0.495, 0.495, 0.165, 0.165, 0.495, 0.495, 0.165]
|
209 |
+
- model: Sao10K/L3-8B-Lunaris-v1
|
210 |
+
parameters:
|
211 |
+
density: [0.65, 0.55, 0.5, 0.45, 0.35, 0.45, 0.5, 0.55, 0.65]
|
212 |
+
weight: [0.495, 0.165, 0.165, 0.495, 0.495, 0.165, 0.165, 0.495]
|
213 |
+
merge_method: dare_ties
|
214 |
+
base_model: nothingiisreal/L3-8B-Celeste-v1
|
215 |
+
parameters:
|
216 |
+
normalize: false
|
217 |
+
int8_mask: true
|
218 |
+
dtype: float32
|
219 |
+
out_dtype: bfloat16
|
220 |
```
|
221 |
|
222 |
***
|
223 |
### Super-Nova-UNC_pt.1
|
224 |
|
225 |
```yaml
|
226 |
+
models:
|
227 |
+
- model: turboderp/llama3-turbcat-instruct-8b
|
228 |
+
- model: ChaoticNeutrals/Domain-Fusion-L3-8B
|
229 |
+
parameters:
|
230 |
+
density: 0.5
|
231 |
+
weight: [0.495, 0.165, 0.165, 0.495, 0.495, 0.165, 0.165, 0.495]
|
232 |
+
- model: migtissera/Llama-3-8B-Synthia-v3.5
|
233 |
+
parameters:
|
234 |
+
density: 0.5
|
235 |
+
weight: [0.165, 0.495, 0.495, 0.165, 0.165, 0.495, 0.495, 0.165]
|
236 |
+
merge_method: dare_ties
|
237 |
+
base_model: turboderp/llama3-turbcat-instruct-8b
|
238 |
+
parameters:
|
239 |
+
normalize: false
|
240 |
+
int8_mask: true
|
241 |
+
dtype: float32
|
242 |
+
out_dtype: bfloat16
|
243 |
```
|
244 |
|
245 |
***
|
246 |
### Super-Nova-UNC_pt.2
|
247 |
|
248 |
```yaml
|
249 |
+
models:
|
250 |
+
- model: turboderp/llama3-turbcat-instruct-8b
|
251 |
+
- model: TheDrummer/Llama-3SOME-8B-v2
|
252 |
+
parameters:
|
253 |
+
density: 0.5
|
254 |
+
weight: [0.165, 0.495, 0.495, 0.165, 0.165, 0.495, 0.495, 0.165]
|
255 |
+
- model: ChaoticNeutrals/Hathor_RP-v.01-L3-8B
|
256 |
+
parameters:
|
257 |
+
density: 0.5
|
258 |
+
weight: [0.495, 0.165, 0.165, 0.495, 0.495, 0.165, 0.165, 0.495]
|
259 |
+
merge_method: dare_ties
|
260 |
+
base_model: turboderp/llama3-turbcat-instruct-8b
|
261 |
+
parameters:
|
262 |
+
normalize: false
|
263 |
+
int8_mask: true
|
264 |
+
dtype: float32
|
265 |
+
out_dtype: bfloat16
|
266 |
```
|
267 |
|
268 |
***
|
269 |
### Super-Nova-INT_pt.1
|
270 |
|
271 |
```yaml
|
272 |
+
models:
|
273 |
+
- model: TheSkullery/llama-3-cat-8b-instruct-v1
|
274 |
+
- model: FPHam/L3-8B-Everything-COT
|
275 |
+
parameters:
|
276 |
+
density: 0.5
|
277 |
+
weight: [0.139, 0.139, 0.208, 0.139, 0.208]
|
278 |
+
- model: Ayush-1722/Meta-Llama-3-8B-Instruct-Summarize-v0.2-24K-LoRANET-Merged
|
279 |
+
parameters:
|
280 |
+
density: 0.5
|
281 |
+
weight: [0.139, 0.208, 0.139, 0.208, 0.139]
|
282 |
+
- model: OEvortex/Emotional-llama-8B
|
283 |
+
parameters:
|
284 |
+
density: 0.5
|
285 |
+
weight: [0.208, 0.139, 0.208, 0.139, 0.139]
|
286 |
+
- model: lighteternal/Llama3-merge-biomed-8b
|
287 |
+
parameters:
|
288 |
+
density: 0.5
|
289 |
+
weight: [0.208, 0.139, 0.139, 0.139, 0.208]
|
290 |
+
- model: Casual-Autopsy/Llama3-merge-psychotherapy-8b
|
291 |
+
parameters:
|
292 |
+
density: 0.5
|
293 |
+
weight: [0.139, 0.208, 0.139, 0.208, 0.139]
|
294 |
+
merge_method: ties
|
295 |
+
base_model: TheSkullery/llama-3-cat-8b-instruct-v1
|
296 |
+
parameters:
|
297 |
+
normalize: false
|
298 |
+
int8_mask: true
|
299 |
+
dtype: float32
|
300 |
+
out_dtype: bfloat16
|
301 |
```
|
302 |
|
303 |
***
|
304 |
### Super-Nova-INT_pt.2
|
305 |
|
306 |
```yaml
|
307 |
+
models:
|
308 |
+
- model: TheSkullery/llama-3-cat-8b-instruct-v1
|
309 |
+
- model: FPHam/L3-8B-Everything-COT
|
310 |
+
parameters:
|
311 |
+
density: 0.9
|
312 |
+
gamma: 0.01
|
313 |
+
weight: [0.139, 0.208, 0.208, 0.139, 0.139]
|
314 |
+
- model: Ayush-1722/Meta-Llama-3-8B-Instruct-Summarize-v0.2-24K-LoRANET-Merged
|
315 |
+
parameters:
|
316 |
+
density: 0.9
|
317 |
+
gamma: 0.01
|
318 |
+
weight: [0.208, 0.139, 0.139, 0.139, 0.208]
|
319 |
+
- model: OEvortex/Emotional-llama-8B
|
320 |
+
parameters:
|
321 |
+
density: 0.9
|
322 |
+
gamma: 0.01
|
323 |
+
weight: [0.139, 0.139, 0.208, 0.208, 0.139]
|
324 |
+
- model: lighteternal/Llama3-merge-biomed-8b
|
325 |
+
parameters:
|
326 |
+
density: 0.9
|
327 |
+
gamma: 0.01
|
328 |
+
weight: [0.139, 0.208, 0.139, 0.208, 0.139]
|
329 |
+
- model: Casual-Autopsy/Llama3-merge-psychotherapy-8b
|
330 |
+
parameters:
|
331 |
+
density: 0.9
|
332 |
+
gamma: 0.01
|
333 |
+
weight: [0.208, 0.139, 0.139, 0.139, 0.208]
|
334 |
+
merge_method: breadcrumbs_ties
|
335 |
+
base_model: TheSkullery/llama-3-cat-8b-instruct-v1
|
336 |
+
parameters:
|
337 |
+
normalize: false
|
338 |
+
int8_mask: true
|
339 |
+
dtype: float32
|
340 |
+
out_dtype: bfloat16
|
341 |
```
|
342 |
|
343 |
***
|
|
|
345 |
|
346 |
```yaml
|
347 |
|
348 |
+
models:
|
349 |
+
- model: Casual-Autopsy/Super-Nova-CRE_pt.1
|
350 |
+
- model: Casual-Autopsy/Super-Nova-CRE_pt.2
|
351 |
+
merge_method: slerp
|
352 |
+
base_model: Casual-Autopsy/Super-Nova-CRE_pt.1
|
353 |
+
parameters:
|
354 |
+
t:
|
355 |
+
- filter: self_attn
|
356 |
+
value: [0.5, 0.3, 0.7, 0.5, 0.7, 0.3, 0.5, 0.3, 0.7, 0.5, 0.7, 0.3, 0.5]
|
357 |
+
- filter: mlp
|
358 |
+
value: [0.5, 0.7, 0.3, 0.5, 0.3, 0.7, 0.5, 0.7, 0.3, 0.5, 0.3, 0.7, 0.5]
|
359 |
+
- value: 0.5
|
360 |
+
embed_slerp: true
|
361 |
+
dtype: float32
|
362 |
+
out_dtype: bfloat16
|
363 |
```
|
364 |
|
365 |
***
|
366 |
### Super-Nova-UNC
|
367 |
|
368 |
```yaml
|
369 |
+
models:
|
370 |
+
- model: Casual-Autopsy/Super-Nova-UNC_pt.1
|
371 |
+
- model: Casual-Autopsy/Super-Nova-UNC_pt.2
|
372 |
+
merge_method: slerp
|
373 |
+
base_model: Casual-Autopsy/Super-Nova-UNC_pt.1
|
374 |
+
parameters:
|
375 |
+
t:
|
376 |
+
- value: [0.5, 0.7, 0.3, 0.5, 0.3, 0.7, 0.5, 0.7, 0.3, 0.5, 0.3, 0.7, 0.5]
|
377 |
+
embed_slerp: true
|
378 |
+
dtype: float32
|
379 |
+
out_dtype: bfloat16
|
380 |
```
|
381 |
|
382 |
***
|
383 |
### Super-Nova-INT
|
384 |
|
385 |
+
```yaml
|
386 |
+
models:
|
387 |
+
- model: Casual-Autopsy/Super-Nova-INT_pt.1
|
388 |
+
- model: Casual-Autopsy/Super-Nova-INT_pt.2
|
389 |
+
merge_method: slerp
|
390 |
+
base_model: Casual-Autopsy/Super-Nova-INT_pt.1
|
391 |
+
parameters:
|
392 |
+
t:
|
393 |
+
- value: 0.5
|
394 |
+
embed_slerp: true
|
395 |
+
dtype: float32
|
396 |
+
out_dtype: bfloat16
|
397 |
+
```
|
398 |
+
|
399 |
+
***
|
400 |
+
### Super-Nova-RP_stp.1
|
401 |
+
|
402 |
```yaml
|
403 |
|
404 |
+
models:
|
405 |
+
- model: Casual-Autopsy/Super-Nova-CRE
|
406 |
+
- model: asual-Autopsy/Super-Nova-UNC
|
407 |
+
merge_method: slerp
|
408 |
+
base_model: Casual-Autopsy/Super-Nova-CRE
|
409 |
+
parameters:
|
410 |
+
t:
|
411 |
+
- value: [0.7, 0.5, 0.3, 0.25, 0.2, 0.25, 0.3, 0.5, 0.7]
|
412 |
+
embed_slerp: true
|
413 |
+
dtype: float32
|
414 |
+
out_dtype: bfloat16
|
415 |
+
|
416 |
```
|
417 |
|
418 |
***
|
419 |
+
### Super-Nova-RP_stp.2
|
420 |
|
421 |
```yaml
|
422 |
+
models:
|
423 |
+
- model: Casual-Autopsy/Super-Nova-RP_stp.1
|
424 |
+
- model: Casual-Autopsy/Super-Nova-INT
|
425 |
+
merge_method: slerp
|
426 |
+
base_model: Casual-Autopsy/Super-Nova-RP_stp.1
|
427 |
+
parameters:
|
428 |
+
t:
|
429 |
+
- value: [0.1, 0.15, 0.2, 0.4, 0.6, 0.4, 0.2, 0.15, 0.1]
|
430 |
+
embed_slerp: true
|
431 |
+
dtype: float32
|
432 |
+
out_dtype: bfloat16
|
433 |
+
```
|
434 |
|
435 |
+
***
|
436 |
+
### Super-Nova-RP_pt.1
|
437 |
+
|
438 |
+
```yaml
|
439 |
+
models:
|
440 |
+
- model: Casual-Autopsy/Super-Nova-RP_stp.2
|
441 |
+
- model: Sao10K/L3-8B-Tamamo-v1
|
442 |
+
parameters:
|
443 |
+
density: [0.4, 0.6, 0.5, 0.6, 0.4]
|
444 |
+
epsilon: [0.15, 0.15, 0.25, 0.15, 0.15]
|
445 |
+
lambda: 0.85
|
446 |
+
weight: [-0.01523, 0.01768, -0.01384, 0.01835, -0.01247]
|
447 |
+
- model: ResplendentAI/Nymph_8B
|
448 |
+
parameters:
|
449 |
+
density: [0.65, 0.35, 0.5, 0.35, 0.65]
|
450 |
+
epsilon: [0.1, 0.1, 0.25, 0.1, 0.1]
|
451 |
+
lambda: 0.85
|
452 |
+
weight: [0.01823, -0.01647, 0.01422, -0.01975, 0.01128]
|
453 |
+
- model: ChaoticNeutrals/T-900-8B
|
454 |
+
parameters:
|
455 |
+
density: [0.35, 0.65, 0.5, 0.65, 0.35]
|
456 |
+
epsilon: [0.1, 0.1, 0.25, 0.1, 0.1]
|
457 |
+
lambda: 0.85
|
458 |
+
weight: [-0.01891, 0.01554, -0.01325, 0.01791, -0.01458]
|
459 |
+
- model: Sao10K/L3-8B-Niitama-v1
|
460 |
+
parameters:
|
461 |
+
density: [0.6, 0.4, 0.5, 0.4, 0.6]
|
462 |
+
epsilon: [0.15, 0.15, 0.25, 0.15, 0.15]
|
463 |
+
lambda: 0.85
|
464 |
+
weight: [0.01768, -0.01675, 0.01285, -0.01696, 0.01421]
|
465 |
+
merge_method: della
|
466 |
+
base_model: Casual-Autopsy/Super-Nova-RP_stp.2
|
467 |
+
parameters:
|
468 |
+
normalize: false
|
469 |
+
int8_mask: true
|
470 |
+
dtype: float32
|
471 |
+
out_dtype: bfloat16
|
472 |
```
|
473 |
|
474 |
***
|
475 |
### Super-Nova-RP_pt.2
|
476 |
|
477 |
```yaml
|
478 |
+
models:
|
479 |
+
- model: Casual-Autopsy/Super-Nova-RP_stp.2
|
480 |
+
- model: bluuwhale/L3-SthenoMaidBlackroot-8B-V1
|
481 |
+
parameters:
|
482 |
+
density: [0.4, 0.6, 0.5, 0.6, 0.4]
|
483 |
+
epsilon: [0.15, 0.15, 0.25, 0.15, 0.15]
|
484 |
+
lambda: 0.85
|
485 |
+
weight: [-0.01935, 0.01785, -0.01512, 0.01809, -0.01371]
|
486 |
+
- model: Hastagaras/Jamet-8B-L3-MK.V-Blackroot
|
487 |
+
parameters:
|
488 |
+
density: [0.65, 0.35, 0.5, 0.35, 0.65]
|
489 |
+
epsilon: [0.1, 0.1, 0.25, 0.1, 0.1]
|
490 |
+
lambda: 0.85
|
491 |
+
weight: [0.01847, -0.01468, 0.01503, -0.01822, 0.01459]
|
492 |
+
- model: Hastagaras/Halu-8B-Llama3-Blackroot
|
493 |
+
parameters:
|
494 |
+
density: [0.35, 0.65, 0.5, 0.65, 0.35]
|
495 |
+
epsilon: [0.1, 0.1, 0.25, 0.1, 0.1]
|
496 |
+
lambda: 0.85
|
497 |
+
weight: [-0.01578, 0.01821, -0.01753, 0.01677, -0.01442]
|
498 |
+
- model: crestf411/L3-8B-sunfall-v0.4-stheno-v3.2
|
499 |
+
parameters:
|
500 |
+
density: [0.6, 0.5, 0.5, 0.5, 0.6]
|
501 |
+
epsilon: [0.15, 0.15, 0.25, 0.15, 0.15]
|
502 |
+
lambda: 0.85
|
503 |
+
weight: [0.01667, -0.01740, 0.01560, -0.01564, 0.01315]
|
504 |
+
merge_method: della
|
505 |
+
base_model: Casual-Autopsy/Super-Nova-RP_stp.2
|
506 |
+
parameters:
|
507 |
+
normalize: false
|
508 |
+
int8_mask: true
|
509 |
+
dtype: float32
|
510 |
+
out_dtype: bfloat16
|
511 |
```
|
512 |
|
513 |
***
|
514 |
### L3-Super-Nova-RP-8B
|
515 |
|
516 |
```yaml
|
517 |
+
models:
|
518 |
+
- model: Casual-Autopsy/Super-Nova-RP_stp.2
|
519 |
+
- model: /kaggle/input/super-nova-rp_pt.4/transformers/hf/1
|
520 |
+
merge_method: slerp
|
521 |
+
base_model: /kaggle/input/super-nova-rp_pt.3/transformers/hf/1
|
522 |
+
parameters:
|
523 |
+
t:
|
524 |
+
- value: 0.5
|
525 |
+
dtype: float32
|
526 |
+
out_dtype: bfloat16
|
527 |
```
|