INDEX

Explanations

maximum values or limits specified in a context

oai_token-act-pair · gpt-3.5-turbo

references to the concept of "maximum" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

stuff

-0.69

 lately

-0.67

rust

-0.66

born

-0.63

ware

-0.63

rum

-0.61

talk

-0.61

flies

-0.61

beat

-0.61

love

-0.60

POSITIVE LOGITS

 maximum

3.68

maximum

2.72

 Maximum

2.46

Maximum

2.17

 minimum

2.07

 maximal

2.02

max

1.97

 optimum

1.81

 Minimum

1.60

minimum

1.60

Activations Density 0.012%

maximum values or limits specified in a context

references to the concept of "maximum" in various contexts

No Comments

No Known Activations