INDEX

Explanations

words related to written communication such as "letter to"

oai_token-act-pair · gpt-3.5-turbo

the occurrence of the phrase "letter to" indicating communication or requests made in letters

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

zees

-0.69

 continuum

-0.67

acles

-0.67

redits

-0.66

avascript

-0.64

igmatic

-0.64

gal

-0.63

entimes

-0.63

ibilities

-0.63

tions

-0.63

POSITIVE LOGITS

 accompany

1.03

 superiors

0.99

 shareholders

0.95

 subscribers

0.93

 editors

0.89

 constituents

0.85

 colleagues

0.83

 commemorate

0.80

 subordinates

0.79

 inbox

0.78

Activations Density 0.123%

words related to written communication such as "letter to"

the occurrence of the phrase "letter to" indicating communication or requests made in letters

No Comments

No Known Activations