INDEX

Explanations

names of positions or roles in a conversation

oai_token-act-pair · gpt-3.5-turbo

references to individuals involved in a conversation or interview context

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

rams

-0.74

 Catholicism

-0.71

 Pradesh

-0.68

 Farming

-0.68

illet

-0.66

ital

-0.64

eanor

-0.64

 Tour

-0.63

 Revival

-0.63

 pilgrimage

-0.63

POSITIVE LOGITS

 laureate

0.85

 misunderstood

0.77

 surn

0.77

NetMessage

0.77

itatively

0.76

who

0.76

 extraord

0.75

 beware

0.74

NAME

0.73

 reviewers

0.71

Activations Density 0.064%

names of positions or roles in a conversation

references to individuals involved in a conversation or interview context

No Comments

No Known Activations