INDEX

Explanations

phrases related to being physically impacted or attacked, often with negative outcomes

oai_token-act-pair · gpt-3.5-turbo

phrases indicating actions or events that are associated with being affected or impacted

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

spection

-0.80

taboola

-0.80

nces

-0.73

ĨĴ

-0.72

SPONSORED

-0.70

ruary

-0.69

theless

-0.68

sylv

-0.67

iltr

-0.67

inois

-0.66

POSITIVE LOGITS

ritic

0.80

 tails

0.70

 lightning

0.69

 snag

0.69

henko

0.67

 runoff

0.66

 plateau

0.65

 stride

0.65

crazy

0.65

 missiles

0.65

Activations Density 0.468%

phrases related to being physically impacted or attacked, often with negative outcomes

phrases indicating actions or events that are associated with being affected or impacted

No Comments

No Known Activations