INDEX

Explanations

The neuron is looking for acronyms consisting of the letters "TH" followed by a number

oai_token-act-pair · gpt-3.5-turbo

occurrences of the acronym "TH" and its variations in different contexts

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

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

ãĤŃ

-0.71

ãĤ±

-0.69

detail

-0.69

assetsadobe

-0.67

cloth

-0.64

mens

-0.64

ãĤ¹ãĥĪ

-0.63

 Canaver

-0.63

atories

-0.62

enda

-0.62

POSITIVE LOGITS

ulhu

1.06

irteen

0.93

ttp

0.92

IRD

0.91

urst

0.86

OUGH

0.86

OTAL

0.84

orne

0.83

omas

0.81

TH

0.81

Activations Density 0.008%

The neuron is looking for acronyms consisting of the letters "TH" followed by a number

occurrences of the acronym "TH" and its variations in different contexts

No Comments

No Known Activations