INDEX

Explanations

sentence endings, possibly focused on conclusions or summaries

oai_token-act-pair · gpt-3.5-turbo

sentences that conclude or summarize discussions

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

isbury

-0.75

undai

-0.73

 negoti

-0.73

 electr

-0.72

 bonded

-0.72

 eleph

-0.72

 displacement

-0.71

 hatch

-0.70

 unthinkable

-0.69

 tremend

-0.69

POSITIVE LOGITS

 Anyway

1.66

Anyway

1.62

 Nonetheless

1.49

 Nevertheless

1.43

 Suff

1.40

 Regardless

1.30

 Hopefully

1.25

 However

1.25

Firstly

1.21

 Readers

1.20

Activations Density 0.404%

sentence endings, possibly focused on conclusions or summaries

sentences that conclude or summarize discussions

No Comments

No Known Activations