INDEX

Explanations

objects or tools that are manually held and operated

oai_token-act-pair · gpt-3.5-turbo

mentions of the word "handle" 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

 circulation

-0.73

issance

-0.69

egal

-0.69

ovych

-0.67

 indo

-0.65

 recomm

-0.64

DEF

-0.63

hemat

-0.61

iary

-0.60

IFIED

-0.60

POSITIVE LOGITS

bars

1.67

bar

1.07

 handle

0.91

 handles

0.89

shake

0.82

tops

0.79

tails

0.79

handle

0.79

pipe

0.78

washer

0.78

Activations Density 0.020%

objects or tools that are manually held and operated

mentions of the word "handle" in various contexts

No Comments

No Known Activations