INDEX

Explanations

action-related terms or calls for intervention

oai_token-act-pair · gpt-3.5-turbo

references to taking action or the concept of action

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

conservancy

-0.81

 abund

-0.69

 Province

-0.63

iciency

-0.63

Kou

-0.62

Fey

-0.62

inately

-0.61

 illusions

-0.61

 abundant

-0.61

ringe

-0.60

POSITIVE LOGITS

ional

0.93

 action

0.87

 Replay

0.87

iveness

0.84

ality

0.82

ivism

0.81

igraph

0.81

 ACTIONS

0.79

uations

0.79

uary

0.77

Activations Density 0.034%

action-related terms or calls for intervention

references to taking action or the concept of action

No Comments

No Known Activations