INDEX

Explanations

references to military positions or high-ranking officials

oai_token-act-pair · gpt-3.5-turbo

repeated phrases and references to titles, particularly involving 'of'

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

yip

-0.76

izable

-0.70

enos

-0.68

rish

-0.64

hunter

-0.64

icans

-0.60

nih

-0.60

hunt

-0.59

rative

-0.59

xon

-0.58

POSITIVE LOGITS

 Circle

0.65

®

0.63

pton

0.62

 Pyth

0.61

edIn

0.60

aiden

0.60

Circ

0.59

 Gaal

0.59

pez

0.58

Connector

0.58

Activations Density 0.145%

references to military positions or high-ranking officials

repeated phrases and references to titles, particularly involving 'of'

No Comments

No Known Activations