INDEX

Explanations

though to a lesser extent, the neuron also identifies government-related terms

oai_token-act-pair · gpt-3.5-turbo

terms related to political content and discussions

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

scl

-0.88

avorite

-0.72

duino

-0.72

gypt

-0.68

manship

-0.68

 practicable

-0.68

nir

-0.66

hire

-0.66

HOME

-0.65

ylum

-0.65

POSITIVE LOGITS

ized

1.62

ization

1.57

ised

1.30

izing

1.30

izations

1.27

isation

1.23

izes

1.22

ize

1.13

ified

1.10

ation

1.05

Activations Density 0.036%

though to a lesser extent, the neuron also identifies government-related terms

terms related to political content and discussions

No Comments

No Known Activations