INDEX

Explanations

the word "rogue" in various contexts

oai_token-act-pair · gpt-3.5-turbo

references to "rogue" entities or groups

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

atsu

-0.97

drops

-0.80

ieu

-0.76

ori

-0.75

arel

-0.73

urity

-0.72

iences

-0.71

enance

-0.71

opped

-0.71

need

-0.71

POSITIVE LOGITS

 rogue

0.85

esses

0.82

Rogue

0.75

 trooper

0.74

^^^^

0.70

 Trooper

0.70

 Rogue

0.66

 troop

0.66

blade

0.64

 Trader

0.64

Activations Density 0.016%

the word "rogue" in various contexts

references to "rogue" entities or groups

No Comments

No Known Activations