INDEX

Explanations

phrases with the word "said" or "says" indicating reported speech

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "said" and its variants

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

²¾

-0.87

à¦

-0.77

ATURE

-0.70

aque

-0.69

venth

-0.68

estern

-0.67

atos

-0.67

LET

-0.67

ptives

-0.66

otin

-0.66

POSITIVE LOGITS

 sources

0.85

 analysts

0.80

 according

0.75

 spokeswoman

0.74

 critics

0.73

 Sheila

0.72

 spokesman

0.71

 Olson

0.70

 organizers

0.70

 analyst

0.69

Activations Density 0.101%

phrases with the word "said" or "says" indicating reported speech

occurrences of the word "said" and its variants

No Comments

No Known Activations