Parameters van LLM
Net als elke machine learningmodel, hebben grote taalmodellen diverse parameters die de variantie van de gegenereerde tekstuitvoer controleren. We hebben een meerdelijkrij-serie begonnen om de impact van deze parameters in detail uit te leggen. We zullen afsluiten door de perfecte balans in contentgeneratie te vinden, gebruik makende van alle deze parameters die we in ons meerdelijkrij-series hebben besproken.
Welkom bij de tweedeel, waar we over een ander bekend parameter, “Top-P,” praten.
Top-P (Nucleus Sampling)
Als het doel is de diversiteit van de modeluitvoer te controleren, is Top-P de gekozen parameter. Een lage Top-P dwingt het model om de meest waarschijnlijke woorden te gebruiken, terwijl een hoge Top-P de model dwingt om meer diverse woorden te gebruiken, waardoor de creativiteit toeneemt.
Laten we Top-P in actie zien met behulp van het volgende code en uitvoer.
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))
Uitvoer:
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.
Laten we nu de uitvoer begrijpen.
- Top-P 0.1 – heel conservatief:Omdat het model kiest uit de top 10% van de waarschijnlijke volgende woordkeuzes, is er veel herhaling in de gegenereerde inhoud. Daarom ontbreekt deze reactie aan diversiteit en is ze vaak ook niet informatief.
- Top-P 0.3 – Conservatief: Het model kiest uit de top 30% van de waarschijnlijke volgende woordkeuzes, dus het is iets minder conservatief dan het vorige Top-P instelling.als u uit de uitvoer ziet, is dit niet verbeterd contentgeneratie en de prompt werd door elkaar gehaald in de voltooiing. In dit geval betekent de herhaling van de prompt de meest waarschijnlijke vervolging na de prompt voor het model lijkt zichzelf te zijn.
- Top-P 0.5 – Balanсed: Dit is waar u voor het eerst de model lijst wat getalde strategieën ziet. U ziet nog steeds wat herhaling in deze instelling ook. Maar het spijt is dat bij deze Top-P instelling, het model beginnen om een bredere keuze van woorden te integreren. De uitvoer is een mengsel van standaard raadvoorstellen met enkele on一致heden. Deze Top-P waarde staat toe om creativiteit te verbeteren, maar heeft nog steeds moeite met de diepte van de informatie.
- Top-P 0.7 – Creatief: In dit geval kan het model een bredere keuze van woorden maken, en zoals u kunt zien, verschuift de reactie naar een narratieve stijl. Het content is meer creatief omdat het nu een scenario bevat waarin iemand met stress omgaat. Het nadeel is de verlies van focus, aangezien de nadruk niet op het beheersen van stress lag, maar op de moeilijkheden om met stress om te gaan.
- Top-P 0.9 – heel Creatief: In deze instelling heeft het model toegang tot een breed scala aan vocabulaire en ideeën inclusief minder waarschijnlijke woorden en concepten. Deze instelling heeft het model geïnspireerd om meer expressieve taal te gebruiken. Opnieuw het nadeel van heel creatief zijn dat de model afwijkt van de prompt in de zoektocht naar het produceren van rijke en variantie content.
Het belangrijkste uit de bovenstaande oefening is hoe de inhoud verandert met de verandering in de Top-P instelling. Het geeft ons ook een idee dat dit parameter niet het enige is dat behandeld moet worden voor variantie in de inhoud en haar relevantie.
Nu laten we kijken wat de impact is van Top-P op een paar gebruiksgevallen, net zoals in de vorige onderdeel van deze reeks over “Creatieve Verhaal Generatie” en “Technische Explicatie.”
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
Uitvoer:
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."
Nu laten we de uitvoer voor creatieve verhaalgeneratie en technische verklaring analyseren op basis van de Top-P instellingen en hoe de uitvoer beïnvloed werd.
Om de impact van Top-P effectief te demonstreren, hebben we betere prompts ingepast om de uitvoer in de goede richting te sturen zodat het effect gemakkelijk waargenomen kan worden.
Creatieve Verhaal Generatie
- Laag Top-P (Negatieve Impact): Zoals u kunt zien met het lage Top-P, is de model beperkt tot het gebruik van woorden of frases, waardoor herhaling en redundantie ontstaan. De creativiteit is ook beperkt in dit geval, aangezien het model probeert geen nieuwe ideeën te introduceren. Maar merk op dat de logische stroom nog steeds worden behouden en het model blijft op het onderwerp, wat kenmerkend is voor lagere Top-P waarden.
- Hoog Top-P (Perfect Impact): In dit geval introduceert de model nieuwe concepten en voegt een creatieve hoek toe aan de vertelling. Bredere woordenbank wordt gebruikt, die diepte en rijkdom aan de tekst toevoegt. Helaas is de logische stroom, door de vergroot creativiteit, beperkt.
De contrast tussen de twee narraties laat duidelijk zien de impact van Top-P, waardoor het gemakkelijk te begrijpen is hoe het effect heeft op creatieve schrijven.
technische uitleg
- Hoge Top-P (Negatieve impact):Als u kunt zien heeft een hoge Top-P een negatief effect op technische uitleg door het verhinderen van een logische stroom en afwijken van het onderwerp. Het model introduceert ook irrelevante informatie die niet aan de uitleg relevant is.
- Optimaal Top-P (Perfecte impact):De uitleg is meer coherent en dicht bij het onderwerp met een optimale Top-P. Het inhoudelijke aligneert beter met de opdracht en balanceert nauwkeurigheid en uitdrukking goed. De betrouwbaarheid van de informatie wordt verhoogd omdat het model beperkt is tot meer waarschijnlijke woorden.
Conclusie
Met deze experimenten hebben we de belangrijkheid van de Top-P parameter getoond in het controleren van de willekeurigheid en creativiteit van de gegenereerde tekst. We hebben eerst een enkele opdracht gezien en hoe de uitvoer verschilt met verschillende Top-P en daarna een meer case-gebonden aanpak gevolgd hoe Top-P de uitvoer controleert op basis van het gebruiksdoel.
Hieruit en uit de vorige delen van de reeks kunnen we echter zien dat elke parameter afzonderlijk niet voldoende voor de kwaliteit van de inhoud genereren. Dat is waarom het essentieel is om het effect van alleParameters te bekijken, en we zullen dat doen als het laatste deel van deze reeks.
Source:
https://dzone.com/articles/decoding-llm-parameters-top-p