INDEX

Explanations

words related to negative connotations, criticism, or derogatory language

oai_token-act-pair · gpt-3.5-turbo

words related to meaning and interpretations of actions or statements

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

è¦ļéĨĴ

-0.72

ideo

-0.71

Witness

-0.69

ãĥīãĥ©ãĤ´ãĥ³

-0.67

 Blaz

-0.66

Miller

-0.66

 GEAR

-0.65

 Jail

-0.65

 Mesa

-0.64

 Hurricanes

-0.64

POSITIVE LOGITS

aning

1.03

roth

0.82

roman

0.82

atos

0.78

rowth

0.75

theless

0.75

etheless

0.75

awan

0.74

onymous

0.73

 effic

0.72

Activations Density 0.010%

words related to negative connotations, criticism, or derogatory language

words related to meaning and interpretations of actions or statements

No Comments

No Known Activations