INDEX

Explanations

phrases related to dialogue or interviews, specifically with a conversational tone

oai_token-act-pair · gpt-3.5-turbo

expressions of dialogue or conversational cues

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

"!

-0.70

 Slug

-0.69

 Osaka

-0.67

otech

-0.65

 upkeep

-0.65

!".

-0.65

ãĤĬ

-0.64

 relegation

-0.62

 discarded

-0.61

pur

-0.60

POSITIVE LOGITS

Secretary

0.81

NPR

0.75

 Sabha

0.74

 Congressman

0.73

——

0.73

 GOODMAN

0.72

 Secretary

0.71

rien

0.71

President

0.70

udeau

0.68

Activations Density 0.479%

phrases related to dialogue or interviews, specifically with a conversational tone

expressions of dialogue or conversational cues

No Comments

No Known Activations