INDEX

Explanations

words related to communication or speaking

oai_token-act-pair · gpt-3.5-turbo

instances of human communication or speech

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

Crew

-0.73

edia

-0.71

 defaults

-0.71

ĨĴ

-0.69

 References

-0.68

secondary

-0.67

dash

-0.66

 Slug

-0.66

iris

-0.65

docs

-0.65

POSITIVE LOGITS

iannopoulos

0.75

zai

0.74

np

0.69

antes

0.69

uterte

0.68

 Archdemon

0.67

jriwal

0.67

ao

0.66

opez

0.65

abbit

0.64

Activations Density 0.000%

words related to communication or speaking

instances of human communication or speech

No Comments

No Known Activations