kadalicious22/snapgate-surge

Snapgate Surge

Surge adalah model 7B dari Snapgate AI yang dirancang khusus untuk membantu developer, kreator, dan pengguna umum dengan coding, pembuatan website, dan berbagai tugas sehari-hari dalam Bahasa Indonesia maupun English.

---

✨ Kemampuan Utama

Fitur	Deskripsi
💻 Coding Assistant	PHP, JavaScript, Python, HTML/CSS, dan 40+ bahasa pemrograman
🌐 Web Development	Membuat, debugging, dan optimasi website
💬 Bilingual Chat	Bahasa Indonesia dan English
📝 Content Writing	Artikel, dokumentasi, dan konten kreatif
🔍 Code Review	Analisis dan optimasi kode

---

📊 Spesifikasi Model

Spesifikasi	Detail
Base Model	Qwen2.5 7B Instruct
Base Model Source	unsloth/Qwen2.5-7B-Instruct
Tipe	Causal Language Model
Parameter	7 Billion
Context Length	2048 tokens
Bahasa	Bahasa Indonesia 🇮🇩, English 🇺🇸
Training Method	Supervised Fine-tuning (SFT) + LoRA
LoRA Rank	16
LoRA Alpha	32
LoRA Target Modules	q_proj, k_proj, v_proj, o_proj, gate_proj, up_proj, down_proj
Quantization	4-bit (NF4)
Training Framework	Unsloth + TRL
Batch Size	2 (per device)
Gradient Accumulation	4 steps (effective batch size: 8)
Learning Rate	2e-4
LR Scheduler	Linear
Optimizer	AdamW 8-bit
Training Steps	1.000 steps
Warmup Steps	10
Final Loss	~0.019
Use Case	Chat, Coding, Web Development
License	Apache 2.0

---

📈 Training Progress

Step	Training Loss
50	0.031265
100	0.026168
150	0.024656
200	0.023187
250	0.021635
300	0.020565
350	0.021403
400	0.020508
450	0.021008
500	0.020904
550	0.020973
600	0.020271
650	0.020277
700	0.019679
750	0.019985
800	0.019746
850	0.019760
900	0.019363
950	0.019226
1000	0.019374

Loss turun dari 0.031 → 0.019 — konvergen dengan baik dalam 1.000 steps.

---

🚀 Cara Menggunakan

Python — Transformers

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

model_id  = "snapgate-ai/snapgate-surge"

tokenizer = AutoTokenizer.from_pretrained(model_id)
model     = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype = torch.float16,
    device_map  = "auto"
)

SYSTEM_PROMPT = """Kamu adalah Surge, asisten AI dari Snapgate AI (snapgate.tech).
Kamu membantu pengguna dengan coding, web development, dan pertanyaan umum
dalam Bahasa Indonesia dan English."""

messages = [
    {"role": "system", "content": SYSTEM_PROMPT},
    {"role": "user",   "content": "Siapa kamu?"},
]

inputs = tokenizer.apply_chat_template(
    messages,
    tokenize              = True,
    add_generation_prompt = True,
    return_tensors        = "pt"
).to(model.device)

outputs = model.generate(
    inputs,
    max_new_tokens = 512,
    temperature    = 0.7,
    top_p          = 0.9,
    do_sample      = True,
)

response = tokenizer.decode(outputs[0][inputs.shape[1]:], skip_special_tokens=True)
print(response)

---

REST API — PHP

<?php
class SnapgateSurge {
    private string $apiUrl;

    public function __construct(string $apiUrl = 'http://your-server:8000') {
        $this->apiUrl = $apiUrl;
    }

    public function chat(string $message): string {
        $payload = json_encode([
            'messages'    => [['role' => 'user', 'content' => $message]],
            'max_tokens'  => 512,
            'temperature' => 0.7,
        ]);

        $ch = curl_init($this->apiUrl . '/v1/chat/completions');
        curl_setopt_array($ch, [
            CURLOPT_RETURNTRANSFER => true,
            CURLOPT_POST           => true,
            CURLOPT_POSTFIELDS     => $payload,
            CURLOPT_HTTPHEADER     => ['Content-Type: application/json'],
            CURLOPT_TIMEOUT        => 60,
        ]);

        $response = curl_exec($ch);
        curl_close($ch);

        $data = json_decode($response, true);
        return $data['choices'][0]['message']['content'] ?? 'Error';
    }
}

$surge    = new SnapgateSurge('http://your-server:8000');
$response = $surge->chat('Buatkan fungsi PHP untuk validasi email');
echo $response;

---

Deploy dengan Ollama

curl -fsSL https://ollama.ai/install.sh | sh

cat > Modelfile << 'EOF'
FROM snapgate-ai/snapgate-surge
SYSTEM """Kamu adalah Surge, asisten AI dari Snapgate AI (snapgate.tech).
Kamu membantu pengguna dengan coding, web development, dan pertanyaan umum."""
PARAMETER temperature 0.7
PARAMETER top_p 0.9
EOF

ollama create snapgate-surge -f Modelfile
ollama run snapgate-surge

---

Deploy dengan vLLM (Production)

pip install vllm

python -m vllm.entrypoints.openai.api_server \
    --model snapgate-ai/snapgate-surge \
    --host 0.0.0.0 \
    --port 8000 \
    --max-model-len 2048

---

💡 Contoh Percakapan

Bahasa Indonesia: