INDEX

Explanations

references to people's actions or activities, particularly involving going somewhere or doing something

oai_token-act-pair · gpt-3.5-turbo

instances of the verb "went."

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

ament

-0.77

uctor

-0.75

eers

-0.72

icio

-0.71

 Horus

-0.70

ullah

-0.69

ribution

-0.68

ipation

-0.68

itionally

-0.67

creen

-0.66

POSITIVE LOGITS

Ń·

0.97

vt

0.94

lems

0.84

XL

0.82

 overboard

0.82

©¶æ

0.79

 ahead

0.79

 unnoticed

0.78

 viral

0.77

":"/

0.77

Activations Density 0.082%

references to people's actions or activities, particularly involving going somewhere or doing something

instances of the verb "went."

No Comments

No Known Activations