INDEX

Explanations

phrases containing the word "arg"

oai_token-act-pair · gpt-3.5-turbo

references to specific authors or contributors within a context

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

FORMATION

-0.86

PER

-0.69

 spaced

-0.68

missions

-0.66

DAY

-0.65

VID

-0.65

 tremend

-0.63

mission

-0.63

spect

-0.61

fulness

-0.60

POSITIVE LOGITS

uably

1.31

emouth

1.09

uments

1.07

ansas

1.07

entin

1.00

raph

0.99

roup

0.99

irl

0.95

eering

0.92

allery

0.92

Activations Density 0.029%

phrases containing the word "arg"

references to specific authors or contributors within a context

No Comments

No Known Activations