INDEX

Explanations

phrases with strong negative connotations or criticism

oai_token-act-pair · gpt-3.5-turbo

instances of the word "blatant" and its variants to identify clear examples of misinformation or authority misuse

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

ulton

-0.80

otos

-0.76

arios

-0.75

nesota

-0.75

iership

-0.75

adal

-0.74

 safely

-0.74

agine

-0.72

oan

-0.71

oult

-0.70

POSITIVE LOGITS

 misrepresent

0.85

 violation

0.84

iary

0.82

ãĥĨ

0.81

 violations

0.80

 omission

0.78

 disregard

0.77

 contradiction

0.75

faced

0.74

 blatant

0.74

Activations Density 0.020%

phrases with strong negative connotations or criticism

instances of the word "blatant" and its variants to identify clear examples of misinformation or authority misuse

No Comments

No Known Activations