INDEX

Explanations

phrases related to reporting actions or incidents

oai_token-act-pair · gpt-3.5-turbo

instances of political discourse or themes related to authenticity and representation

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 downed

-0.77

NetMessage

-0.72

 Sabha

-0.71

 subsequ

-0.71

asus

-0.67

onel

-0.65

 Bought

-0.63

 Sergeant

-0.63

 sergeant

-0.61

watching

-0.61

POSITIVE LOGITS

 realism

1.02

traditional

0.96

 eclectic

0.96

 populist

0.93

 inclusive

0.93

 multicultural

0.92

 orthodox

0.88

 unorthodox

0.88

 traditional

0.86

 overtly

0.86

Activations Density 1.394%

phrases related to reporting actions or incidents

instances of political discourse or themes related to authenticity and representation

No Comments

No Known Activations