INDEX

Explanations

phrases related to communication, particularly expressing opinions or initiating dialogue

oai_token-act-pair · gpt-3.5-turbo

emotive expressions or exclamations

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.10.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

senal

-0.90

 citiz

-0.88

aditional

-0.85

quartered

-0.84

awei

-0.80

 eleph

-0.79

 favoured

-0.76

 satell

-0.73

 pione

-0.72

 teasp

-0.72

POSITIVE LOGITS

Seriously

0.94

—-

0.86

lol

0.85

Yep

0.82

ï¸ı

0.82

Honestly

0.80

Í

0.77

 âĿ

0.76

âĶĢâĶĢâĶĢâĶĢ

0.75

ÂłÂłÂł

0.75

Activations Density 0.302%

phrases related to communication, particularly expressing opinions or initiating dialogue

emotive expressions or exclamations

No Comments

No Known Activations