INDEX

Explanations

references to significant news events or incidents

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

phrases related to news and reporting

oai_token-act-pair · gpt-3.5-turbo

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 attrition

-0.88

 ecology

-0.88

 mapping

-0.87

 parity

-0.87

 dart

-0.86

 totem

-0.86

 elves

-0.86

 aven

-0.85

 snowball

-0.85

 marqu

-0.84

POSITIVE LOGITS

 Them

1.46

 Shows

1.44

 Claims

1.44

 Needs

1.41

 Called

1.40

 Says

1.40

 Coverage

1.40

 Away

1.40

 Possible

1.38

 Costs

1.38

Activations Density 0.745%

references to significant news events or incidents

phrases related to news and reporting

No Comments

No Known Activations