INDEX

Explanations

data analysis related words and phrases

oai_token-act-pair · gpt-3.5-turbo

phrases related to polling data and survey results

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

ascus

-0.74

via

-0.73

 sidx

-0.70

ufact

-0.69

scribe

-0.69

EEE

-0.65

 ãĤµãĥ¼ãĥĨãĤ£ãĥ¯ãĥ³

-0.65

Synopsis

-0.64

inventoryQuantity

-0.64

uristic

-0.64

POSITIVE LOGITS

 millennials

0.76

 otherwise

0.73

 majorities

0.73

 homelessness

0.72

 median

0.71

 that

0.71

 nobody

0.71

 none

0.70

 respondents

0.70

 viewership

0.67

Activations Density 0.129%

data analysis related words and phrases

phrases related to polling data and survey results

No Comments

No Known Activations