INDEX

Explanations

words related to medical conditions, especially the adjective "sick" with varying intensities

oai_token-act-pair · gpt-3.5-turbo

references to the word "sick" 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 @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

 unlaw

-0.78

rul

-0.71

 compr

-0.69

merce

-0.67

 unden

-0.65

 sanctioned

-0.65

 Unch

-0.65

ETHOD

-0.63

 principals

-0.61

NPR

-0.61

POSITIVE LOGITS

ening

1.35

ened

1.26

bay

1.16

nesses

0.98

er

0.93

ly

0.92

ness

0.90

le

0.88

ert

0.87

igan

0.87

Activations Density 0.021%

words related to medical conditions, especially the adjective "sick" with varying intensities

references to the word "sick" in various contexts

No Comments

No Known Activations