INDEX

Explanations

information or actions that have been reported by a source

oai_token-act-pair · gpt-3.5-turbo

phrases that emphasize the word "reportedly," indicating information that is attributed to sources or claims rather than being confirmed facts

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Appearances

-0.87

bern

-0.85

ERAL

-0.82

uit

-0.76

vation

-0.72

itionally

-0.71

 surely

-0.71

CHECK

-0.70

batch

-0.69

egg

-0.69

POSITIVE LOGITS

 overheard

0.79

 instructed

0.74

 priced

0.73

 obtained

0.72

 originated

0.70

 entertained

0.69

 influenced

0.69

 violated

0.69

 Marketable

0.69

 proc

0.69

Activations Density 0.052%

information or actions that have been reported by a source

phrases that emphasize the word "reportedly," indicating information that is attributed to sources or claims rather than being confirmed facts

No Comments

No Known Activations