INDEX

Explanations

descriptions of how something appears or is observed

oai_token-act-pair · gpt-3.5-turbo

instances of the word "looked" and its variations in different contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

arta

-0.78

ency

-0.73

[+]

-0.69

been

-0.68

ibel

-0.68

wall

-0.67

opal

-0.67

tan

-0.67

por

-0.64

arel

-0.63

POSITIVE LOGITS

nesday

0.90

 tremend

0.82

 UCHIJ

0.76

hes

0.76

cffffcc

0.76

hers

0.74

 initially

0.70

 originally

0.69

 teasp

0.69

 unsuccessfully

0.69

Activations Density 0.276%

descriptions of how something appears or is observed

instances of the word "looked" and its variations in different contexts

No Comments

No Known Activations