INDEX

Explanations

content related to systematic actions or processes

oai_token-act-pair · gpt-3.5-turbo

references to systematic processes or actions

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

gets

-0.82

nery

-0.79

tein

-0.78

hawk

-0.77

hawks

-0.76

boards

-0.74

frog

-0.73

jon

-0.73

bolt

-0.73

busters

-0.72

POSITIVE LOGITS

 systematic

0.92

itarian

0.87

urally

0.81

izational

0.80

 overhaul

0.79

 economical

0.78

 violation

0.78

icity

0.77

 numerical

0.75

analysis

0.75

Activations Density 0.024%

content related to systematic actions or processes

references to systematic processes or actions

No Comments

No Known Activations