INDEX

Explanations

dates written in non-standard format

oai_token-act-pair · gpt-3.5-turbo

occurrences of numerical data or significant dates in a narrative context

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

enaries

-0.68

 himself

-0.64

 dissu

-0.63

 answ

-0.59

ĸļ

-0.59

 disadvant

-0.58

 themselves

-0.58

 posit

-0.58

}}}

-0.58

 Tsukuyomi

-0.58

POSITIVE LOGITS

IRO

0.77

ccording

0.76

Warning

0.76

Known

0.75

OUT

0.74

croft

0.74

NEW

0.73

LOS

0.73

HOU

0.73

Contact

0.70

Activations Density 0.433%

dates written in non-standard format

occurrences of numerical data or significant dates in a narrative context

No Comments

No Known Activations