INDEX

Explanations

the word "the" when followed by a specific type of word or context

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "the" along with various objects or entities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

amar

-0.83

isan

-0.82

thood

-0.82

forestation

-0.81

WAYS

-0.81

incinnati

-0.76

ruary

-0.76

hov

-0.73

ablishment

-0.73

ornings

-0.71

POSITIVE LOGITS

 offending

1.26

 accelerator

1.19

lid

1.17

 brakes

1.10

 railing

1.10

 trigger

1.09

 ignition

1.08

 handcuffs

1.07

 knob

1.07

 steering

1.06

Activations Density 0.143%

the word "the" when followed by a specific type of word or context

occurrences of the word "the" along with various objects or entities

No Comments

No Known Activations