INDEX

Explanations

terms related to a location or direction, particularly the front or facing direction

oai_token-act-pair · gpt-3.5-turbo

the term "front" 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 @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

ATURE

-0.74

 mortals

-0.67

ibl

-0.64

 Duncan

-0.62

Ģ

-0.62

 curse

-0.60

 typing

-0.60

 shar

-0.59

iv

-0.58

 stringent

-0.57

POSITIVE LOGITS

front

4.51

Front

2.50

 fronts

1.84

 Front

1.83

 front

1.66

ront

1.24

side

1.11

back

1.02

facing

1.01

forward

1.00

Activations Density 0.010%

terms related to a location or direction, particularly the front or facing direction

the term "front" in various contexts

No Comments

No Known Activations