INDEX

Explanations

references to something mentioned previously in a text

oai_token-act-pair · gpt-3.5-turbo

references to previously mentioned information or subjects in the text

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

CTR

-0.74

 Picks

-0.67

 Poles

-0.67

 Guard

-0.65

Mail

-0.64

Ring

-0.64

 Markets

-0.64

pocket

-0.63

UNCH

-0.63

thumbnails

-0.63

POSITIVE LOGITS

mentioned

1.30

 mentioned

0.93

ground

0.88

above

0.76

 equation

0.75

 applies

0.74

 snippet

0.74

ebin

0.72

inance

0.71

nder

0.71

Activations Density 0.029%

references to something mentioned previously in a text

references to previously mentioned information or subjects in the text

No Comments

No Known Activations