Update README.md
Browse files
README.md
CHANGED
|
@@ -1,16 +1,76 @@
|
|
| 1 |
---
|
| 2 |
-
title:
|
| 3 |
-
emoji:
|
| 4 |
-
colorFrom:
|
| 5 |
-
colorTo:
|
| 6 |
sdk: gradio
|
| 7 |
sdk_version: 5.33.0
|
| 8 |
app_file: app.py
|
| 9 |
pinned: false
|
| 10 |
license: mit
|
| 11 |
-
short_description: '
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12 |
---
|
| 13 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 14 |
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
|
| 15 |
|
| 16 |
# π Instagram Caption AI Model Benchmark
|
|
@@ -19,12 +79,12 @@ This benchmark evaluates **Caption Generation** and **Multi-Language Translation
|
|
| 19 |
|
| 20 |
## π― Caption Generation Models
|
| 21 |
|
| 22 |
-
| Model ID | Provider | Avg Latency | Caption Quality | Multi-Modal |
|
| 23 |
-
|
| 24 |
-
| `Llama-4-Maverick-17B-128E` π | SambaNova | **2.1s** | **Excellent** | β
Yes |
|
| 25 |
-
| `GPT-4-Vision` | OpenAI | 3.2s | Excellent | β
Yes |
|
| 26 |
-
| `Claude-3-Vision` | Anthropic | 2.8s | Very Good | β
Yes |
|
| 27 |
-
| `Gemini-Pro-Vision` | Google | 2.5s | Good | β
Yes |
|
| 28 |
|
| 29 |
**β
Chosen Primary Model:** `Llama-4-Maverick-17B-128E-Instruct`
|
| 30 |
- **Instagram-specialized prompting** with hashtag optimization
|
|
@@ -34,12 +94,12 @@ This benchmark evaluates **Caption Generation** and **Multi-Language Translation
|
|
| 34 |
|
| 35 |
## β¨ Caption Variation Models
|
| 36 |
|
| 37 |
-
| Model ID | Provider | Avg Latency | Variation Quality |
|
| 38 |
-
|
| 39 |
-
| `Meta-Llama-3.2-3B` π | SambaNova | **1.4s** | **Excellent** |
|
| 40 |
-
| `GPT-3.5-Turbo` | OpenAI | 2.1s | Good |
|
| 41 |
-
| `Claude-3-Haiku` | Anthropic | 1.8s | Very Good |
|
| 42 |
-
| `Gemma-2-9B` | Google | 1.6s | Good |
|
| 43 |
|
| 44 |
**β
Chosen Variation Model:** `Meta-Llama-3.2-3B-Instruct`
|
| 45 |
- **3 distinct approaches:** Story-driven, Question-based, Value-packed
|
|
|
|
| 1 |
---
|
| 2 |
+
title: Instagram Caption AI Studio
|
| 3 |
+
emoji: π±
|
| 4 |
+
colorFrom: purple
|
| 5 |
+
colorTo: pink
|
| 6 |
sdk: gradio
|
| 7 |
sdk_version: 5.33.0
|
| 8 |
app_file: app.py
|
| 9 |
pinned: false
|
| 10 |
license: mit
|
| 11 |
+
short_description: 'AI-Powered Instagram Caption Generator'
|
| 12 |
+
tags:
|
| 13 |
+
- instagram
|
| 14 |
+
- caption-generator
|
| 15 |
+
- sambanova
|
| 16 |
+
- llama
|
| 17 |
+
- multi-language
|
| 18 |
+
- huggingface
|
| 19 |
+
- social-media
|
| 20 |
+
- ai
|
| 21 |
+
- computer-vision
|
| 22 |
+
- translation
|
| 23 |
+
- content-creation
|
| 24 |
+
- viral-marketing
|
| 25 |
+
suggested_hardware: cpu-basic
|
| 26 |
+
suggested_storage: small
|
| 27 |
+
models:
|
| 28 |
+
- SambaNova/Llama-4-Maverick-17B-128E-Instruct
|
| 29 |
+
- Meta-Llama-3.2-3B-Instruct
|
| 30 |
+
- google-t5/t5-small
|
| 31 |
+
- chence08/mt5-small-iwslt2017-zh-en
|
| 32 |
+
- Helsinki-NLP/opus-mt-en-hi
|
| 33 |
+
- marefa-nlp/marefa-mt-en-ar
|
| 34 |
+
|
| 35 |
+
language:
|
| 36 |
+
- en
|
| 37 |
+
- de
|
| 38 |
+
- zh
|
| 39 |
+
- hi
|
| 40 |
+
- ar
|
| 41 |
+
library_name: gradio
|
| 42 |
+
base_path: /
|
| 43 |
+
custom_headers:
|
| 44 |
+
cross-origin-embedder-policy: require-corp
|
| 45 |
+
cross-origin-opener-policy: same-origin
|
| 46 |
---
|
| 47 |
|
| 48 |
+
# π± Instagram Caption AI Studio
|
| 49 |
+
|
| 50 |
+
> π **Advanced AI-Powered Instagram Content Creation Suite**
|
| 51 |
+
|
| 52 |
+
## β¨ Key Features
|
| 53 |
+
|
| 54 |
+
π€ **SambaNova Integration**: Llama-4-Maverick + Llama-3.2-3B models
|
| 55 |
+
π **Multi-Language**: German, Chinese, Hindi, Arabic translation
|
| 56 |
+
πΌοΈ **Vision AI**: Multi-modal image analysis with quality scoring
|
| 57 |
+
π― **Smart Targeting**: 8 caption styles Γ 8 audience types
|
| 58 |
+
β¨ **Variations**: Generate 3 alternative captions instantly
|
| 59 |
+
π₯ **Instagram Optimized**: Hashtag generation & engagement prediction
|
| 60 |
+
|
| 61 |
+
## π οΈ Technology Stack
|
| 62 |
+
|
| 63 |
+
- **Primary AI**: SambaNova Llama-4-Maverick-17B-128E-Instruct
|
| 64 |
+
- **Variations**: Meta-Llama-3.2-3B-Instruct
|
| 65 |
+
- **Translation**: Hugging Face T5, MT5, Helsinki-NLP, Marefa models
|
| 66 |
+
- **Interface**: Advanced Gradio with custom glassmorphism UI
|
| 67 |
+
- **Performance**: <2.1s caption generation, <1.4s variations
|
| 68 |
+
|
| 69 |
+
## π― Perfect For
|
| 70 |
+
|
| 71 |
+
Content creators, social media managers, influencers, brands, and anyone looking to create engaging Instagram content with AI assistance.
|
| 72 |
+
|
| 73 |
+
**Try it now and create viral-worthy captions in seconds!** π
|
| 74 |
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
|
| 75 |
|
| 76 |
# π Instagram Caption AI Model Benchmark
|
|
|
|
| 79 |
|
| 80 |
## π― Caption Generation Models
|
| 81 |
|
| 82 |
+
| Model ID | Provider | Avg Latency | Caption Quality | Multi-Modal |
|
| 83 |
+
|-----------------------------------|-------------|-------------|-----------------|-------------|
|
| 84 |
+
| `Llama-4-Maverick-17B-128E` π | SambaNova | **2.1s** | **Excellent** | β
Yes |
|
| 85 |
+
| `GPT-4-Vision` | OpenAI | 3.2s | Excellent | β
Yes |
|
| 86 |
+
| `Claude-3-Vision` | Anthropic | 2.8s | Very Good | β
Yes |
|
| 87 |
+
| `Gemini-Pro-Vision` | Google | 2.5s | Good | β
Yes |
|
| 88 |
|
| 89 |
**β
Chosen Primary Model:** `Llama-4-Maverick-17B-128E-Instruct`
|
| 90 |
- **Instagram-specialized prompting** with hashtag optimization
|
|
|
|
| 94 |
|
| 95 |
## β¨ Caption Variation Models
|
| 96 |
|
| 97 |
+
| Model ID | Provider | Avg Latency | Variation Quality |
|
| 98 |
+
|-----------------------------|-------------|-------------|-------------------|
|
| 99 |
+
| `Meta-Llama-3.2-3B` π | SambaNova | **1.4s** | **Excellent** |
|
| 100 |
+
| `GPT-3.5-Turbo` | OpenAI | 2.1s | Good |
|
| 101 |
+
| `Claude-3-Haiku` | Anthropic | 1.8s | Very Good |
|
| 102 |
+
| `Gemma-2-9B` | Google | 1.6s | Good |
|
| 103 |
|
| 104 |
**β
Chosen Variation Model:** `Meta-Llama-3.2-3B-Instruct`
|
| 105 |
- **3 distinct approaches:** Story-driven, Question-based, Value-packed
|