INDEX

Explanations

words related to proper names, potentially full names of individuals

oai_token-act-pair · gpt-3.5-turbo

occurrences of the sequence of letters 'us' or similar patterns within words

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

 envy

-0.83

FACE

-0.61

 mathemat

-0.61

ACTED

-0.61

 Curiosity

-0.60

FTWARE

-0.60

ModLoader

-0.59

 resolve

-0.59

 Babel

-0.59

 readiness

-0.59

POSITIVE LOGITS

ich

0.91

nick

0.84

eman

0.84

enberg

0.84

orst

0.83

opoulos

0.83

chuk

0.83

angan

0.83

endor

0.83

akis

0.82

Activations Density 0.315%

words related to proper names, potentially full names of individuals

occurrences of the sequence of letters 'us' or similar patterns within words

No Comments

No Known Activations