INDEX

Explanations

references to speeches or public addresses made by individuals

oai_token-act-pair · gpt-3.5-turbo

instances of the word "speech."

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

olen

-0.71

rome

-0.67

 Trout

-0.66

involved

-0.65

 Popular

-0.64

astern

-0.63

rafted

-0.61

 Soldiers

-0.61

olved

-0.61

 Associ

-0.61

POSITIVE LOGITS

 speech

1.04

speech

0.92

anooga

0.88

 Speech

0.87

 speeches

0.87

writer

0.81

writers

0.81

 uttered

0.78

 speaker

0.78

 imped

0.76

Activations Density 0.022%

references to speeches or public addresses made by individuals

instances of the word "speech."

No Comments

No Known Activations