INDEX

Explanations

concerns or expressions of worry within texts

oai_token-act-pair · gpt-3.5-turbo

references to worries or issues

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

nice

-0.71

ctors

-0.71

gall

-0.69

buff

-0.67

graph

-0.64

SW

-0.62

INAL

-0.62

cle

-0.61

Interview

-0.60

tiny

-0.60

POSITIVE LOGITS

afety

1.06

 concerns

0.91

 regarding

0.85

 Concern

0.81

 raised

0.80

 pertaining

0.79

 relating

0.78

hooting

0.78

 concern

0.78

 arising

0.76

Activations Density 0.030%

concerns or expressions of worry within texts

references to worries or issues

No Comments

No Known Activations