INDEX

Explanations

personal pronouns followed by verbs expressing an action or a state

oai_token-act-pair · gpt-3.5-turbo

common phrases and questions directed at the audience or reader

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

onto

-0.72

arget

-0.67

ologies

-0.67

Draw

-0.67

arta

-0.67

encies

-0.67

requires

-0.66

Adds

-0.65

etc

-0.65

Neither

-0.65

POSITIVE LOGITS

 originally

0.89

 last

0.86

 conceived

0.81

 instrumental

0.79

 gracious

0.78

 ãĤµãĥ¼ãĥĨãĤ£ãĥ¯ãĥ³

0.77

 yesterday

0.74

KGB

0.70

 revelation

0.70

 initially

0.68

Activations Density 0.843%

personal pronouns followed by verbs expressing an action or a state

common phrases and questions directed at the audience or reader

No Comments

No Known Activations