GGUF
Finnish
File size: 2,685 Bytes
e14b0f3
 
 
 
 
 
89794e5
 
0a35c12
 
755bf60
0a35c12
 
 
 
 
755bf60
 
ecd0b81
755bf60
79752e0
 
c048fc5
755bf60
0a35c12
 
f06be02
b6734e9
f53771a
79752e0
 
0a35c12
58c92da
0a35c12
3c06ced
 
f53771a
0a35c12
f06be02
b6734e9
f53771a
aa3e8b6
f53771a
0a35c12
 
 
 
 
 
 
 
 
 
 
f06be02
0a35c12
b605852
aa3e8b6
0a35c12
 
58c92da
0a35c12
3c06ced
 
b605852
0a35c12
f06be02
b6734e9
b605852
576aaa5
b605852
0a35c12
 
 
576aaa5
 
aa3e8b6
576aaa5
4f9631c
576aaa5
6e6c80c
576aaa5
 
 
aa3e8b6
576aaa5
 
 
 
 
 
f06be02
576aaa5
b605852
576aaa5
 
 
 
aa3e8b6
b605852
aa3e8b6
f06be02
576aaa5
b605852
aa3e8b6
576aaa5
b605852
0a35c12
b605852
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
---
license: apache-2.0
language:
- fi
base_model:
- LumiOpen/Viking-7B
datasets:
- ArttuPakarinen/AI-chat-Summarization-FIN
---

Full fine-tune on LumiOpen's viking 7B, 4bit kvantization - for testing.

Around 3-4 million tokens full fine tune to make the model produce summarization.

Around 2-6 hours of compute on A100 GPU.

Partial dataset provided on the model card - other datasets used were Diagsum and Samsum, which were autotranslated to finnish.

Model works fine for it's purpose, even in this low-parameter and low kvantization mode.

Model has been trained to start "first prompt" with "Ihminen aloittaa keskustelun AI-botin kanssa\n\n". I call this 'First summanry'. Without adding the first summanry, some weird behaviour may occur.

**Examples**

*Example1*

***INPUT***

```text
Ihminen aloittaa keskustelun AI-botin kanssa.

Ihminen: Hei mitä kuuluu?

AI-botti: Hyvää kuuluu, kuinka voi palvellla?

###TIIVISTELMÄ###
```

***OUTPUT***

```text
Ihminen ja AI-botti tervehtivät toisiaan ja AI-botti kysyy kuinka voi palvella.
```



---------------

The idea is to summarize the chat context, and to be able to continue chatting, by using the produced summarization in the begigning of the next prompt.

---------------

*Example2*

***INPUT***

```text
Ihminen ja AI-botti tervehtivät toisiaan ja AI-botti kysyy kuinka voi palvella.

Ihminen: Kuinka paljon ihmisiä Suomessa asuu?

AI-botti: Suomessa asuu n. 5,5 miljoonaa ihmistä.

###TIIVISTELMÄ###
```

***OUTPUT***

```text
Ihminen ja AI-botti tervehtivät toisiaan, jonka jälkeen ihminen kysyy AI-botilta paljonko Suomessa asuu ihmisiä. AI-botti vastasi, että n. 5,5 miljoonaa ihmistä.
```



---------------

In theory, the model can be used to produce AI responses, but the model will always produce 

\#\#\#TIIVISTELMÄ\#\#\#

\<summarization\>



This means, if you want to produce AI answer, it is completely possible by ending the prompt with "\nAI-botti: ", but you need to parse the summarization part away from the response.


---------------

*Example3*

***INPUT***

```text
Ihminen ja AI-botti tervehtivät toisiaan, jonka jälkeen ihminen kysyy AI-botilta paljonko Suomessa asuu ihmisiä. AI-botti vastasi, että n. 5,5 miljoonaa ihmistä.

Ihminen: Ahaa, mutta entäs Ruotsissa?

AI-botti: 
```

***OUTPUT***

```text
Ruotsissa asuu lähes tuplasti väkeä Suomeen verrattuna, reilut 10 miljoonaa asukasta. 

###TIIVISTELMÄ###

Ihminen kysyy AI-botilta paljonko Suomessa asuu ihmisiä. AI-botti vastasi, että n. 5,5 miljoonaa ihmistä. Ihminen kysyi paljonko Ruotsissa asuu ihmisiä, johon AI-botti vastasi että lähes tuplasti Suomeen verrattuna, päälle 10 miljoonaa ihmistä.
```