INDEX

Explanations

words related to directional indications like "right" or "left"

oai_token-act-pair · gpt-3.5-turbo

references to the concept of "right" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

aton

-0.85

anned

-0.82

eson

-0.78

angan

-0.72

odium

-0.69

Clar

-0.68

atically

-0.68

atis

-0.67

rolet

-0.66

ensable

-0.65

POSITIVE LOGITS

 wing

1.13

 flank

1.00

wing

0.94

eye

0.89

most

0.89

ward

0.86

 hand

0.84

 hemisphere

0.84

 side

0.84

eous

0.82

Activations Density 0.037%

words related to directional indications like "right" or "left"

references to the concept of "right" in various contexts

No Comments

No Known Activations