INDEX

Explanations

names of individuals and positions, possibly related to investigations

oai_token-act-pair · gpt-3.5-turbo

the conjunction "and" used 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 @ 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

asper

-0.76

 incent

-0.74

DEP

-0.63

HQ

-0.62

spot

-0.62

natureconservancy

-0.62

ctica

-0.62

asse

-0.61

heads

-0.60

ios

-0.60

POSITIVE LOGITS

 assorted

0.75

 thence

0.74

romeda

0.71

 consequently

0.70

 secondly

0.69

igham

0.69

 possibly

0.68

 vice

0.67

 optionally

0.67

alus

0.66

Activations Density 0.269%

names of individuals and positions, possibly related to investigations

the conjunction "and" used in various contexts

No Comments

No Known Activations