Upload ./README.md with huggingface_hub
Browse files
README.md
ADDED
@@ -0,0 +1,79 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: apache-2.0
|
3 |
+
datasets:
|
4 |
+
- Epiculous/SynthRP-Gens-v1.1-Filtered-n-Cleaned
|
5 |
+
- anthracite-org/stheno-filtered-v1.1
|
6 |
+
- PJMixers/hieunguyenminh_roleplay-deduped-ShareGPT
|
7 |
+
- Gryphe/Sonnet3.5-Charcard-Roleplay
|
8 |
+
- Epiculous/Synthstruct-Gens-v1.1-Filtered-n-Cleaned
|
9 |
+
- anthracite-org/kalo-opus-instruct-22k-no-refusal
|
10 |
+
- anthracite-org/nopm_claude_writing_fixed
|
11 |
+
- anthracite-org/kalo_opus_misc_240827
|
12 |
+
language:
|
13 |
+
- en
|
14 |
+
- fr
|
15 |
+
- de
|
16 |
+
- es
|
17 |
+
- it
|
18 |
+
- pt
|
19 |
+
- ru
|
20 |
+
- zh
|
21 |
+
- ja
|
22 |
+
pipeline_tag: text-generation
|
23 |
+
tags:
|
24 |
+
- merge
|
25 |
+
---
|
26 |
+
### exl2 quant (measurement.json in main branch)
|
27 |
+
---
|
28 |
+
### check revisions for quants
|
29 |
+
---
|
30 |
+
|
31 |
+
|
32 |
+
![image/png](https://cdn-uploads.huggingface.co/production/uploads/64adfd277b5ff762771e4571/P962FQhRG4I8nbU_DJolY.png)
|
33 |
+
|
34 |
+
Now for something a bit different, Violet_Twilight-v0.2! This model is a SLERP merge of Azure_Dusk-v0.2 and Crimson_Dawn-v0.2!
|
35 |
+
|
36 |
+
# Quants!
|
37 |
+
<strong>full</strong> / [exl2]() / [gguf]()
|
38 |
+
|
39 |
+
## Prompting
|
40 |
+
The v0.2 models are trained on ChatML, the prompting structure goes a little something like this:
|
41 |
+
|
42 |
+
```
|
43 |
+
<|im_start|>user
|
44 |
+
Hi there!<|im_end|>
|
45 |
+
<|im_start|>assistant
|
46 |
+
Nice to meet you!<|im_end|>
|
47 |
+
<|im_start|>user
|
48 |
+
Can I ask a question?<|im_end|>
|
49 |
+
<|im_start|>assistant
|
50 |
+
```
|
51 |
+
|
52 |
+
### Context and Instruct
|
53 |
+
The v0.2 models are trained on ChatML, please use that Context and Instruct template.
|
54 |
+
|
55 |
+
### Current Top Sampler Settings
|
56 |
+
[Spicy_Temp](https://files.catbox.moe/9npj0z.json) <br/>
|
57 |
+
[Violet_Twilight-Nitral-Special](https://files.catbox.moe/ot54u3.json) <br/>
|
58 |
+
|
59 |
+
## Merging
|
60 |
+
The following config was used to merge Azure Dusk and Crimson Dawn
|
61 |
+
```yaml
|
62 |
+
slices:
|
63 |
+
- sources:
|
64 |
+
- model: Epiculous/Azure_Dusk-v0.2
|
65 |
+
layer_range: [0, 40]
|
66 |
+
- model: Epiculous/Crimson_Dawn-V0.2
|
67 |
+
layer_range: [0, 40]
|
68 |
+
merge_method: slerp
|
69 |
+
base_model: Epiculous/Azure_Dusk-v0.2
|
70 |
+
parameters:
|
71 |
+
t:
|
72 |
+
- filter: self_attn
|
73 |
+
value: [0, 0.5, 0.3, 0.7, 1]
|
74 |
+
- filter: mlp
|
75 |
+
value: [1, 0.5, 0.7, 0.3, 0]
|
76 |
+
- value: 0.5 # fallback for rest of tensors
|
77 |
+
dtype: bfloat16
|
78 |
+
|
79 |
+
```
|