INDEX

Explanations

phrases related to actions and capabilities

oai_token-act-pair · gpt-3.5-turbo

phrases indicating conditional situations or consequences

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

raft

-0.77

Â¯

-0.73

mx

-0.71

ode

-0.67

neg

-0.65

owl

-0.64

dds

-0.64

uga

-0.63

sav

-0.63

MSN

-0.62

POSITIVE LOGITS

 accordingly

1.04

 thereafter

0.90

 alike

0.72

 consequently

0.70

 consequ

0.67

 advoc

0.66

 notor

0.66

 reused

0.66

 afterwards

0.65

 thereof

0.64

Activations Density 0.808%

phrases related to actions and capabilities

phrases indicating conditional situations or consequences

No Comments

No Known Activations