INDEX

Explanations

contrasting statements or information

oai_token-act-pair · gpt-3.5-turbo

phrases indicating a sense of denial or expectation regarding events that were supposed to occur

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

Tonight

-0.70

YP

-0.68

ril

-0.68

However

-0.66

si

-0.66

Its

-0.66

tv

-0.65

sic

-0.64

 However

-0.64

Republic

-0.63

POSITIVE LOGITS

etheless

0.78

 nonetheless

0.76

 undeniably

0.73

 balk

0.73

 persuaded

0.73

 remarkably

0.72

 horm

0.71

 dogged

0.70

 exceedingly

0.67

 stubborn

0.66

Activations Density 1.167%

contrasting statements or information

phrases indicating a sense of denial or expectation regarding events that were supposed to occur

No Comments

No Known Activations