INDEX

Explanations

words related to continuity or remaining unchanged

oai_token-act-pair · gpt-3.5-turbo

instances of the word "remain" indicating ongoing states or conditions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

ramid

-0.86

ongyang

-0.70

 Tribune

-0.67

ranch

-0.66

ilar

-0.65

ounter

-0.64

insula

-0.63

 Circuit

-0.62

isson

-0.61

ilateral

-0.61

POSITIVE LOGITS

 unchanged

1.10

 unanswered

0.97

 undecided

0.93

 intact

0.92

 afloat

0.88

 unaffected

0.83

 unsolved

0.83

rences

0.79

 untouched

0.79

nces

0.78

Activations Density 0.023%

words related to continuity or remaining unchanged

instances of the word "remain" indicating ongoing states or conditions

No Comments

No Known Activations