INDEX

Explanations

mentions of specific names, particularly "Edgar" and "Wen"

oai_token-act-pair · gpt-3.5-turbo

references to individuals named Edgar and Wen

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

akia

-0.91

atures

-0.91

tto

-0.88

cloth

-0.85

ton

-0.82

tons

-0.81

ships

-0.79

tics

-0.79

bee

-0.79

body

-0.78

POSITIVE LOGITS

ENC

0.84

ENTS

0.83

ENCY

0.75

 sidx

0.73

avorite

0.71

akura

0.70

ENT

0.69

vernment

0.68

 Edgar

0.68

 tremend

0.67

Activations Density 0.017%

mentions of specific names, particularly "Edgar" and "Wen"

references to individuals named Edgar and Wen

No Comments

No Known Activations