INDEX

Explanations

mentions of specific names, likely related to a legal case or incident

oai_token-act-pair · gpt-3.5-turbo

repeated mentions of the Bundy family

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

oen

-0.76

unction

-0.69

ORED

-0.67

OC

-0.66

ored

-0.66

 Lovely

-0.65

Bok

-0.64

Io

-0.64

ORE

-0.62

nep

-0.61

POSITIVE LOGITS

 Bundy

1.73

 ranch

0.91

anan

0.80

undy

0.79

querque

0.77

inez

0.77

BLM

0.76

illas

0.76

erers

0.75

 ranc

0.75

Activations Density 0.014%

mentions of specific names, likely related to a legal case or incident

repeated mentions of the Bundy family

No Comments

No Known Activations