INDEX

Explanations

dates and numerical values expressed in text

oai_token-act-pair · gpt-3.5-turbo

numerical references, specifically dates and significant numbers

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 behavi

-0.65

theless

-0.63

ogyn

-0.62

 targ

-0.61

 alike

-0.60

abwe

-0.60

 Rica

-0.59

 derog

-0.58

iku

-0.58

 warfare

-0.56

POSITIVE LOGITS

th

1.33

teenth

1.12

venth

1.03

rd

0.99

nd

0.97

%-

0.92

%"

0.88

ieth

0.87

 richest

0.79

ottest

0.79

Activations Density 0.123%

dates and numerical values expressed in text

numerical references, specifically dates and significant numbers

No Comments

No Known Activations