INDEX

Explanations

phrases indicating available or potential space for improvement or growth

oai_token-act-pair · gpt-3.5-turbo

mentions of "room for improvement" in various contexts

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

raid

-0.62

 Unified

-0.62

 Laksh

-0.60

 Silicon

-0.60

Die

-0.58

 discredited

-0.58

 reconnect

-0.58

 Kafka

-0.57

SX

-0.57

NECT

-0.57

POSITIVE LOGITS

room

0.92

iencies

0.81

bos

0.80

orthy

0.79

 room

0.77

ioned

0.76

lain

0.76

dimension

0.74

atile

0.72

bell

0.71

Activations Density 0.019%

phrases indicating available or potential space for improvement or growth

mentions of "room for improvement" in various contexts

No Comments

No Known Activations