INDEX

Explanations

the adjective "apparent."

oai_token-act-pair · gpt-3.5-turbo

the word "apparent" in various contexts

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

uden

-0.87

alez

-0.75

ajo

-0.74

hov

-0.73

toe

-0.72

inion

-0.69

pped

-0.69

 drilled

-0.68

 researched

-0.68

isol

-0.67

POSITIVE LOGITS

 contradictions

0.95

aneously

0.89

 discrepancies

0.80

 contradiction

0.77

iary

0.76

 inconsistency

0.74

 inconsistencies

0.73

aneous

0.73

éĹ

0.72

ibility

0.71

Activations Density 0.006%

the adjective "apparent."

the word "apparent" in various contexts

No Comments

No Known Activations