INDEX

Explanations

words related to referencing or mentioning something specific

oai_token-act-pair · gpt-3.5-turbo

instances of the word "refer" and its variations, indicating discussions or mentions

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

urity

-0.66

hov

-0.56

lite

-0.54

cffff

-0.54

iqueness

-0.54

oppable

-0.53

lishes

-0.53

hold

-0.53

padding

-0.53

lehem

-0.52

POSITIVE LOGITS

to

0.93

 specifically

0.91

rers

0.85

 back

0.75

 explicitly

0.75

 directly

0.74

 sarcast

0.74

 thereto

0.71

 readers

0.70

entious

0.70

Activations Density 0.049%

words related to referencing or mentioning something specific

instances of the word "refer" and its variations, indicating discussions or mentions

No Comments

No Known Activations