INDEX

Explanations

phrases related to toxic substances

oai_token-act-pair · gpt-3.5-turbo

references to toxins or harmful substances

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

noon

-0.88

dar

-0.77

pora

-0.76

aan

-0.69

quarter

-0.66

doms

-0.64

RM

-0.64

Merit

-0.63

romy

-0.63

ann

-0.62

POSITIVE LOGITS

 poisoning

1.07

 poison

1.00

 dart

0.95

 poisoned

0.91

 darts

0.90

iv

0.90

ously

0.85

Ivy

0.84

 poisonous

0.84

gas

0.84

Activations Density 0.012%

phrases related to toxic substances

references to toxins or harmful substances

No Comments

No Known Activations