Hugging Face
Models
Datasets
Spaces
Posts
Docs
Enterprise
Pricing
Log In
Sign Up
127
3
249
Djuunaa
djuna
Follow
John6666's profile picture
ltim's profile picture
ZeroXClem's profile picture
15 followers
·
65 following
AI & ML interests
None yet
Recent Activity
new
activity
about 7 hours ago
dnhkng/RYS-XLarge:
RYS with Qwen2.5
reacted
to
Elizezen
's
post
with 👀
about 12 hours ago
It turned out that the following simple method seems to be actually effective when you want to increase the appearance probability of only one or a very limited number of tokens. ``` import os one_token = "♡" # Token to increase the appearance probability value = 1000000 token = one_token * value with open("one-token.txt", "w", encoding="utf-8") as f: f.write(token) ``` By training LoRA with unsloth based on the .txt file generated by the code above, you can increase the appearance probability of specific tokens while maintaining the model's performance to great extent. However, it's better to stop the training before train loss becomes 0.0, as it will start spamming the token once it appears even once. In general, you can stop training at a very early stage and it will still work. It is also possible to reduce the appearance probability of specific tokens by creating an over-learned LoRA with the specific tokens you want to reduce, combining it with the model, and then creating a model that extracts only the difference using the chat vector method and subtracting it from an arbitrary model. In this case, it is better to set the ratio of chat vector to about five times. It has very little effect on the overall performance, apart from the specific tokens. ``` new_v = v - (5.0 * chat_vector[i].to(v.device)) ```
liked
a model
about 12 hours ago
PocketDoc/Dans-PersonalityEngine-V1.1.0-12b
View all activity
Organizations
djuna
's activity
All
Models
Datasets
Spaces
Papers
Collections
Community
Posts
Upvotes
Likes
liked
a model
about 12 hours ago
PocketDoc/Dans-PersonalityEngine-V1.1.0-12b
Text Generation
•
Updated
5 days ago
•
113
•
12
liked
a model
about 14 hours ago
NarrativAI/Cakrawala-123B
Text Generation
•
Updated
26 days ago
•
199
•
3
liked
a Space
2 days ago
Running
524
📢
UGI Leaderboard
liked
a model
2 days ago
ResplendentAI/Persephone_7B
Text Generation
•
Updated
Mar 17
•
171
•
5
liked
3 models
3 days ago
djuna/Q2.5-Veltha-14B-0.5
Text Generation
•
Updated
3 days ago
•
70
•
7
djuna/Q2.5-Veltha-14B
Text Generation
•
Updated
2 days ago
•
43
•
6
qingy2024/QwQ-14B-Math-v0.2
Text Generation
•
Updated
6 days ago
•
97
•
14
liked
3 models
4 days ago
Delta-Vector/Control-Nanuq-8B
Updated
4 days ago
•
68
•
2
DreadPoor/Casuar-9B-Model_Stock
Text Generation
•
Updated
4 days ago
•
18
•
2
danlou/relay-v0.1-Mistral-Nemo-2407
Text Generation
•
Updated
4 days ago
•
143
•
3
liked
2 models
5 days ago
advokat/AnimePro-FLUX
Text-to-Image
•
Updated
Nov 17
•
462
•
9
maldv/Qwentile2.5-32B-Instruct
Text Generation
•
Updated
5 days ago
•
92
•
7
liked
4 models
6 days ago
KomorebiAI/nllb-200-3.3B-float16-ct2
Translation
•
Updated
6 days ago
•
19
•
3
allura-org/TQ2.5-14B-Neon-v1
Text Generation
•
Updated
3 days ago
•
68
•
5
FallenMerick/Iced-Lemon-Cookie-7B
Text Generation
•
Updated
May 14
•
18
•
1
mpasila/Gemma2-Eclipse-9B
Text Generation
•
Updated
Oct 20
•
20
•
2
liked
2 models
7 days ago
ibm-granite/granite-embedding-107m-multilingual
Sentence Similarity
•
Updated
6 days ago
•
34
•
1
ibm-granite/granite-3.1-8b-instruct
Text Generation
•
Updated
6 days ago
•
4.57k
•
75
liked
a model
9 days ago
KaraKaraWitch/Llama-MiraiFanfare-3.3-70B
Text Generation
•
Updated
10 days ago
•
207
•
6
liked
a Space
10 days ago
Running
3
🌍
Chat Template Tester
Experiment with Chat Template
Load more