INDEX

Explanations

a specific prefix related to planned actions or medical terms

oai_token-act-pair · gpt-3.5-turbo

words related to premeditation or planned actions

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

chnology

-0.71

izons

-0.65

 fundament

-0.64

IRE

-0.64

issues

-0.63

 Luxem

-0.56

ngth

-0.56

WP

-0.54

PK

-0.53

Magazine

-0.53

POSITIVE LOGITS

ieval

1.12

ication

1.08

icative

0.97

icated

0.96

ications

0.95

ãĥ¼ãĤ¯

0.88

iterranean

0.85

achusetts

0.84

icate

0.83

iac

0.81

Activations Density 0.021%

a specific prefix related to planned actions or medical terms

words related to premeditation or planned actions

No Comments

No Known Activations