INDEX

Explanations

phrases indicating something should not be confused with something else

oai_token-act-pair · gpt-3.5-turbo

phrases that include "not to be confused with."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

needs

-0.82

Ļ½

-0.75

 Needs

-0.73

LIN

-0.70

lining

-0.70

¸

-0.70

æ©

-0.68

ãĥ¼ãĥĨãĤ£

-0.66

 Reloaded

-0.65

 Preferences

-0.64

POSITIVE LOGITS

 mention

0.95

 bother

0.91

 worry

0.91

 anymore

0.89

 drown

0.87

 necessarily

0.86

 confuse

0.85

 celebrate

0.83

 exceed

0.80

 disclose

0.79

Activations Density 0.084%

phrases indicating something should not be confused with something else

phrases that include "not to be confused with."

No Comments

No Known Activations