INDEX

Explanations

The main thing this neuron does is find occurrences of the word "barrage"

oai_token-act-pair · gpt-3.5-turbo

terms related to overwhelming attacks or assaults

oai_token-act-pair · gpt-4o-mini Triggered by @bot

the word 'barrage'

oai_token-act-pair · gpt-4-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

OTA

-0.93

uana

-0.80

cius

-0.72

ident

-0.70

occ

-0.69

basketball

-0.68

benef

-0.67

Utah

-0.66

Crime

-0.65

luck

-0.65

POSITIVE LOGITS

 barrage

1.23

 bombard

1.20

 barr

1.11

 inund

1.06

 bombardment

1.05

 pounded

0.93

 onslaught

0.85

geoning

0.80

ãĥ£

0.79

 raged

0.76

Activations Density 0.014%

The main thing this neuron does is find occurrences of the word "barrage"

terms related to overwhelming attacks or assaults

the word 'barrage'

No Comments

No Known Activations