INDEX

Explanations

calls to action prompting the reader to check something out

oai_token-act-pair · gpt-3.5-turbo

call-to-action phrases that encourage checking out content

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

ãĥĨãĤ£

-0.74

 suscept

-0.67

SAY

-0.65

è£ħ

-0.64

Imagine

-0.64

âĹ¼

-0.63

 rendition

-0.62

cffffcc

-0.62

Sin

-0.62

lishes

-0.61

POSITIVE LOGITS

boxes

1.07

mate

1.04

ers

1.00

points

1.00

lists

0.98

list

0.94

hee

0.92

etts

0.90

ups

0.88

enhagen

0.86

Activations Density 0.020%

calls to action prompting the reader to check something out

call-to-action phrases that encourage checking out content

No Comments

No Known Activations