Glaive Code Assistant with Sandboxes

A specialized code assistance model fine-tuned from Qwen3-8B on 131,000 episodes of the glaive-code-assistant-sandboxes dataset. This model excels at code generation, understanding, and sandbox-based execution scenarios.

What This Model Does

This model provides advanced code assistance capabilities with a focus on sandbox execution environments. It has been trained to understand code generation requests, provide context-aware solutions, and work effectively in isolated execution contexts.

**Base Model:** Qwen/Qwen3-8B (8 billion parameters)

**Training Dataset:** DCAgent2/glm46-glaive-code-assistant-sandboxes-maxeps-131k

**License:** Apache 2.0

**HuggingFace ID:** `laion/glm46-glaive-code-assistant-sandboxes-maxeps-131k`

Instructions for AI Agents

When using this model for code assistance tasks, follow these steps:

1. Model Loading and Setup

Load the model using the Transformers library:

```python

from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "laion/glm46-glaive-code-assistant-sandboxes-maxeps-131k"

tokenizer = AutoTokenizer.from_pretrained(model_name)

model = AutoModelForCausalLM.from_pretrained(

model_name,

device_map="auto",

torch_dtype="auto"

)

```

2. Inference Configuration

Use these recommended generation parameters:

```python

generation_config = {

"max_new_tokens": 2048,

"temperature": 0.7,

"top_p": 0.9,

"do_sample": True,

"repetition_penalty": 1.1

}

```

3. Prompt Formatting

Format your prompts to leverage the model's training on conversational code assistance:

```

User: [Your code question or request]