INDEX

Explanations

locations and events, particularly those related to speeches or appearances

oai_token-act-pair · gpt-3.5-turbo

references to events or forums where discussions occur

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

usercontent

-0.83

 sooner

-0.79

 somehow

-0.70

 violates

-0.68

 alive

-0.67

 cheated

-0.66

 prev

-0.64

 altogether

-0.63

 survives

-0.62

 blink

-0.61

POSITIVE LOGITS

 Brookings

0.84

 Carnegie

0.78

 POLITICO

0.77

 commem

0.75

 Methodist

0.75

 Democracy

0.74

 moder

0.72

rium

0.70

 remembrance

0.70

 Emerging

0.69

Activations Density 0.204%

locations and events, particularly those related to speeches or appearances

references to events or forums where discussions occur

No Comments

No Known Activations