Paramètres des GLLM
Comme pour tout modèle d’apprentissage automatique, les grands modèles de langue disposent de divers paramètres qui contrôlent la variance de la sortie de texte générée. Nous avons démarré une série en plusieurs parties pour expliquer en détail l’impact de ces paramètres. Nous conclurons en établissant un équilibre parfait dans la génération de contenu en utilisant tous les paramètres discutés dans notre série en plusieurs parties.
Bienvenue dans la deuxième partie, où nous discutons d’un autre paramètre connu : « Top-P ».
Top-P (Sélection du Noyau)
Si l’objectif est de contrôler la diversité de la sortie du modèle, alors Top-P est l’option idéale pour vous. Un Top-P plus bas force le modèle à utiliser les mots les plus probables, tandis qu’un Top-P plus élevé force le modèle à utiliser des mots plus divers, augmentant la créativité.
Examinons le Top-P en action avec le code et la sortie suivants.
Python
import torch
from transformers import GPT2LMHeadModel, GPT2Tokenizer
# Load GPT-2 model and tokenizer
tokenizer = GPT2Tokenizer.from_pretrained("gpt2")
model = GPT2LMHeadModel.from_pretrained("gpt2")
# Add pad token to tokenizer (GPT-2 doesn't have it by default)
tokenizer.pad_token = tokenizer.eos_token
# Function to generate response with varying top_p
def generate_with_top_p(prompt, top_p):
inputs = tokenizer(prompt, return_tensors='pt', padding=True)
# Set the attention_mask and pad_token_id
outputs = model.generate(
inputs.input_ids,
attention_mask=inputs['attention_mask'],
do_sample=True,
max_length=200,
top_p=top_p,
pad_token_id=tokenizer.eos_token_id
)
return tokenizer.decode(outputs[0], skip_special_tokens=True)
prompt = "What are some effective ways to manage stress in daily life?"
# List of top-p values and their descriptions
top_p_values = {
0.1: "Very conservative: Generates highly probable and safe responses.",
0.3: "Conservative: Generates probable responses with less risk.",
0.5: "Balanced: A mix of safe and creative responses.",
0.7: "Creative: Generates more diverse and creative responses.",
0.9: "Very creative: Allows for highly diverse and less probable responses."
}
# Test top_p variations
for top_p, description in top_p_values.items():
print(f"\nTop-p {top_p} ({description}):\n")
print(generate_with_top_p(prompt, top_p=top_p))
Sortie :
PowerShell
python test_top_p.py
Top-p 0.1 (Very conservative: Generates highly probable and safe responses.):
What are some effective ways to manage stress in daily life?
The following are some of the most common ways to manage stress in daily life.
1. Avoiding stress
The most common way to avoid stress is to avoid it.
The most common way to avoid stress is to avoid it.
2. Avoiding stress
The most common way to avoid stress is to avoid it.
The most common way to avoid stress is to avoid it.
3. Avoiding stress
The most common way to avoid stress is to avoid it.
The most common way to avoid stress is to avoid it.
4. Avoiding stress
The most common way to avoid stress is to avoid it.
The most common way to avoid stress is to avoid it.
5. Avoiding stress
The most common way to avoid stress is to avoid it.
The most common way to avoid stress
Top-p 0.3 (Conservative: Generates probable responses with less risk.):
What are some effective ways to manage stress in daily life?
What are some effective ways to manage stress in daily life?
What are some effective ways to manage stress in daily life?
What are some effective ways to manage stress in daily life?
What are some effective ways to manage stress in daily life?
What are some effective ways to manage stress in daily life?
What are some effective ways to manage stress in daily life?
What are some effective ways to manage stress in daily life?
What are some effective ways to manage stress in daily life?
What are some effective ways to manage stress in daily life?
What are some effective ways to manage stress in daily life?
63
What are some effective ways to manage stress in daily life?
64
65
What are some effective ways to manage stress in daily life?
66
67
What are some effective ways to manage stress in daily life?
68
69
What are some effective
70
71
Top-p 0.5 (Balanced: A mix of safe and creative responses.):
72
73
What are some effective ways to manage stress in daily life?
74
75
1. Stay on top of your body's natural stress levels
76
77
When you're stressed, your body's natural stress levels are low.
78
79
If you're stressed, your body's natural stress levels are high.
80
81
If you're stressed, your body's natural stress levels are low.
82
83
2. Avoid excessive exercise
84
85
Exercise can make you feel better.
86
87
Exercise can make you feel better.
88
89
3. Get up early to avoid fatigue
90
91
Exercise can make you feel better.
92
93
Exercise can make you feel better.
94
95
4. Avoid the temptation to take the wrong thing
96
97
Exercise can make you feel better.
98
99
Exercise can make you feel better.
100
101
5. Avoid eating the wrong foods
102
103
Exercise can make you feel better.
104
105
Exercise can make you feel better.
106
107
6. Avoid the temptation to
108
109
Top-p 0.7 (Creative: Generates more diverse and creative responses.):
110
111
What are some effective ways to manage stress in daily life?
112
113
I am talking about a very specific situation. The person I am talking about has been stressed, but has not been doing much work for a long time. I want to tell you, because this person has had a lot of stress in his life, that it is not something you can just go back to. But what I'm trying to say is, that if you don't have a job, you have to go back to work every day, so you can spend more time with your family. So I've been doing that for a long time now. And so, that is a very common occurrence.
114
115
But what do you think is the best way to deal with the stress?
116
117
You know, it's not easy to deal with it. It is very difficult to deal with the stress that we experience. So, that is a very good way to deal with it. So, I think it's the
118
119
Top-p 0.9 (Very creative: Allows for highly diverse and less probable responses.):
120
121
What are some effective ways to manage stress in daily life?
122
123
There are many things that can be done by daily meditation and practice. As a general rule of thumb, meditation can help you stay mindful of your own needs, goals, feelings, desires, emotions, and the life and emotions around you.
124
125
The purpose of meditation is to feel a deep desire to practice more, to be more mindful, and to be more productive. It also serves to enhance your overall well-being.
126
127
1. Be active, be creative, be mindful, and be optimistic.
128
129
This is where the first step towards meditation comes from. If we're looking for inspiration, there's a whole section on being "active" and "creative."
130
131
While I'm not sure I know much about meditation, I know some of its practitioners and some that I never met. My mom used to tell me that she'd always find a way to make her feel more connected and involved.
Maintenant, essayons de comprendre la sortie.
- Top-P 0.1 – Très conservateur :Étant donné que le modèle sélectionne parmi les 10 % des choix de mot suivant les plus probables, il y a beaucoup de répétitions dans le contenu généré. Par conséquent, cette réponse manque de diversité et est également presque toujours non informatif.
- Top-P 0.3 – Conservateur : Le modèle sélectionne parmi les 30% supérieurs des choix de mot probables suivants, donc il est légèrement moins conservateur que le précédent paramètre Top-P. Comme vous pouvez le voir à l’extrant, cela n’a pas amélioré la génération de contenu, et le prompt a été répété à travers la complétion. Dans ce cas, la répétition du prompt signifie que la suite la plus probable après le prompt pour le modèle semble être le prompt même.
- Top-P 0.5 – Équilibré :C’est ici qu’il apparaît pour la première fois que le modèle énumère certaines stratégies numérotées. Vous voyez toujours une certaine répétition dans ce paramètre également. Mais le point important est que à ce niveau de Top-P, le modèle commence à intégrer un plus large éventail de mots. L’extrant est une combinaison de conseils standards avec quelques incohérences. Cette valeur de Top-P permet une amélioration de la créativité mais continue de souffrir d’une profondeur d’information insuffisante.
- Top-P 0.7 – Créatif : Dans ce cas, le modèle peut sélectionner un plus large éventail de mots, et comme vous pouvez le voir, la réponse est progressivement orientée vers un style narratif. Le contenu est plus créatif car il met maintenant en scène une situation où une personne affronte des stresseurs. Le désavantage est la perte de focus, car l’accent n’est pas mis sur la gestion du stress mais sur les difficultés à faire face au stress.
- Top-P 0.9 – Très Créatif : Dans ce réglage, le modèle a accès à un large éventail de vocabulaire et d’idées, y compris des mots et concepts moins probables. Ce réglage permet au modèle d’utiliser une langue plus expressive. Encore une fois, le désavantage de la créativité excessive est que le modèle dérive du prompt dans sa quête de produire un contenu riche et varié.
Le point majeur à noter à partir de l’exercice précédent est comment le contenu change avec le changement du paramètre Top-P. Il nous donne également l’idée que ce paramètre n’est pas le seul qui doit être traité pour une variation dans le contenu et son pertinence.
Maintenant, regardons l’impact de Top-P sur quelques cas d’utilisation, tout comme dans la partie précédente de cette série sur « La génération de récits créatifs » et « Les explications techniques ».
Python
1
import torch
2
from transformers import GPT2LMHeadModel, GPT2Tokenizer
3
4
# Load GPT-2 model and tokenizer
5
tokenizer = GPT2Tokenizer.from_pretrained("gpt2")
6
model = GPT2LMHeadModel.from_pretrained("gpt2")
7
8
# Add pad token to tokenizer (GPT-2 doesn't have it by default)
9
tokenizer.pad_token = tokenizer.eos_token
10
11
# Function to generate response based on top_p
12
def generate_with_top_p(prompt, top_p, max_length=250):
13
inputs = tokenizer(prompt, return_tensors='pt')
14
outputs = model.generate(
15
inputs.input_ids,
16
attention_mask=inputs.attention_mask,
17
do_sample=True,
18
max_length=max_length,
19
top_p=top_p,
20
pad_token_id=tokenizer.eos_token_id,
21
eos_token_id=tokenizer.eos_token_id,
22
no_repeat_ngram_size=2 # Prevents repetition of phrases
23
)
24
return tokenizer.decode(outputs[0], skip_special_tokens=True)
25
26
### USE CASE 1: CREATIVE STORY GENERATION ###
27
28
def creative_story_generation():
29
prompt = ("In the mystical land of Eldoria, a young warrior found an ancient map "
30
"that led to a hidden treasure guarded by dragons. He knew that courage and "
31
"wisdom would be his allies on this perilous journey.")
32
33
# Negative Impact: Low top_p for creative writing (less creative)
34
print("\n=== Creative Story with Low top_p (0.2) - Negative Impact: ===")
35
low_top_p_story = generate_with_top_p(prompt, top_p=0.2)
36
print(low_top_p_story)
37
38
# Perfect Impact: High top_p for creative writing (more creative)
39
print("\n=== Creative Story with High top_p (0.95) - Perfect Impact: ===")
40
high_top_p_story = generate_with_top_p(prompt, top_p=0.95)
41
print(high_top_p_story)
42
43
### USE CASE 2: TECHNICAL EXPLANATION ###
44
45
def technical_explanation():
46
prompt = ("Explain step by step how the internet works, focusing on how computers "
47
"use IP addresses and data packets to communicate with each other.")
48
49
# Negative Impact: High top_p for technical writing (less precise)
50
print("\n=== Technical Explanation with High top_p (0.95) - Negative Impact: ===")
51
high_top_p_explanation = generate_with_top_p(prompt, top_p=0.95)
52
print(high_top_p_explanation)
53
54
# Perfect Impact: Optimal top_p for technical writing (accurate)
55
print("\n=== Technical Explanation with Optimal top_p (0.5) - Perfect Impact: ===")
56
optimal_top_p_explanation = generate_with_top_p(prompt, top_p=0.5)
57
print(optimal_top_p_explanation)
58
59
# Run both use cases
60
creative_story_generation()
61
technical_explanation()
62
Sortie :
PowerShell
1
python top_p_multiple.py
2
3
=== Creative Story with Low top_p (0.2) - Negative Impact: ===
4
In the mystical land of Eldoria, a young warrior found an ancient map that led to a hidden treasure guarded by dragons. He knew that courage and wisdom would be his allies on this perilous journey.
5
6
The Dragon King
7
...
8
(The Book of the Dragon)
9
,
10
-
11
: The Dragon Lord is a legendary warrior who has been the focus of many legends. The dragon king is the most powerful of all the dragons in the world. In the magical land, he is known as the "Dragon King". He is also known to be the leader of a group of dragons called the Black Dragons. His name is derived from the dragon's name, "the dragon".
12
"The Black Dragon" is an important symbol of power and powerlessness. It is said that the black dragon is able to create a dragon that can defeat the strongest of his enemies. However, the true power of this dragon lies in his ability to manipulate the minds of others. This ability is called "The Dark Dragon". The Dark dragon has a powerful sense of self-preservation and is capable of manipulating others to his will. When he has control over others, his power is so great that he can destroy entire cities. As a result
13
14
=== Creative Story with High top_p (0.95) - Perfect Impact: ===
15
In the mystical land of Eldoria, a young warrior found an ancient map that led to a hidden treasure guarded by dragons. He knew that courage and wisdom would be his allies on this perilous journey.
16
17
Spirits are like gods. In this world, there are no gods without secrets. There are also no secrets about being a fighter or a thief. But every dragon has a special hidden skill, and he or she can use that skill to destroy and gain strength or hide something hidden in the secret. Many dragons are skilled at their martial arts, while most are unaware of the secrets of their true power. These dragons cannot only use these skills, but that will only allow them to escape the dragons' clutches. Because their training will be tested before they're even born, dragon fighting has never been so hard, even without training, so they should be able to break a dragon's body.
18
19
=== Technical Explanation with High top_p (0.95) - Negative Impact: ===
20
Explain step by step how the internet works, focusing on how computers use IP addresses and data packets to communicate with each other. If a person with the same identity as a user on the US government's private network uses the online address bar, then this data is sent to a server on a computer on your local network. Your IP address is a small byte in the string. The IP and network address are identical. Do you remember, you just want to do that instead of using IPs or numbers. In addition, remember that IP can be used to verify a particular IP for you and your computer. For instance, your name does not always match an address on our government network and you should have your public IP in this country. This does seem quite unusual and perhaps a bit bizarre.
21
22
There was a time in Silicon Valley when you could set your identity out. But in most of today's world, how do you set up your own address and how does one look for it? What about the public? The internet itself was different. It was just a set of rules around data flow that you were supposed to follow. Now, even in today the "internet in general" seems a little more complicated to define. Let's say
23
24
=== Technical Explanation with Optimal top_p (0.5) - Perfect Impact: ===
25
Explain step by step how the internet works, focusing on how computers use IP addresses and data packets to communicate with each other.
26
27
"We've been trying to understand how it works and what it means for the future," says James. "It's not just about the IP address, it's about how people communicate. It's also about what's going on with the data. We want to see how this works. What is the Internet going to look like in the next 10 years?"
28
, the director of the Computer Science and Artificial Intelligence Laboratory at the University of Michigan, says that while there's still a lot of work to be done, "we've got to start to think about it."
Maintenant, parsemmons et Analysez la sortie pour la génération de récits créatifs et les explications techniques en fonction des réglages de Top-P et de l’impact sur la sortie.
Pour effectivement démontrer l’impact de Top-P, nous avons intégré des prompts améliorés pour déterminer la sortie de manière à ce que l’impact soit facilement observable.
Génération de récits créatifs
- Top-P bas (Impact négatif) :Comme vous pouvez le voir avec le bas Top-P, le modèle est limité à l’utilisation de mots ou de phrases, ce qui entraîne de la répétition et de la redondance. La créativité est également limitée dans ce cas, car le modèle essaye de ne pas introduire de nouvelles idées. Mais si vous remarquez, la logique de la narration est toujours maintenue et le modèle reste sur le sujet, ce qui est typique de valeurs basses de Top-P.
- Top-P élevé (Impact parfait) :Dans ce cas, le modèle présente de nouveaux concepts et ajoute une composante créative à la narration. Une plus grande variété de vocabulaire est utilisée, ajoutant de la profondeur et de la richesse au texte. Cependant, à cause de la créativité accrue, la logique de la narration a été compromise.
La contraste entre les deux narrations montre clairement l’impact de Top-P, ce qui facilite la compréhension de la manière dont il affecte la rédaction créative.
Explication technique
- Haute Top-P (Impact négatif) : Comme vous pouvez le voir, une haute Top-P a un impact négatif sur les explications techniques en empêchant un flux logique et en dérivant du sujet. Le modèle ajoute également des informations non pertinentes, ce qui n’est pas pertinent à l’explication.
- Top-P optimale (Impact parfait) : Les explications sont plus cohérentes et proches du sujet avec une Top-P optimale. Le contenu se aligne mieux sur le prompt et balance bien l’exactitude et l’expression. La fiabilité des informations est améliorée car le modèle est limité aux mots plus probables.
Conclusion
Avec cette expérience, nous avons réussi à montrer l’importance du paramètre Top-P pour contrôler la randomness et la créativité du texte généré. Nous avons d’abord examiné un seul prompt et comment l’output varie avec différentes valeurs de Top-P, puis adopté une approche plus basée sur les cas d’utilisation pour comprendre comment Top-P contrôle l’output en fonction du cas d’utilisation.
Cependant, à travers les parties précédentes et cette partie de la série, nous avons constaté que individuellement, chaque paramètre ne permet pas une qualité de génération de contenu suffisante. C’est pourquoi il est essentiel de considérer l’impact de tous ces paramètres, ce que nous ferons comme dernière partie de cette série.
Source:
https://dzone.com/articles/decoding-llm-parameters-top-p