The model deepseek-coder-1.3b-typescript is a fine-tuned version of deepseek-ai/deepseek-coder-1.3b-base, specifically optimized for TypeScript code generation. It has been trained on a dataset of 0.5 billion tokens to excel in producing precise and efficient TypeScript solutions. The model supports a 16K window size and includes fill-in-the-middle (FIM) capabilities for project-level code completion.

How to Use

Running the Model on a GPU

from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("CodeGPTPlus/deepseek-coder-1.3b-typescript", trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained("CodeGPTPlus/deepseek-coder-1.3b-typescript", trust_remote_code=True).cuda()

input_text = """function quickSort(arr: number[]): number[] {
  if (arr.length <= 1) {
    return arr;
  }
  const pivot = arr[0];
  const left = [];
  const right = [];
"""

inputs = tokenizer(input_text, return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_length=256)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

Running with Ollama

You can use the model via Ollama by running:

ollama run codegpt/deepseek-coder-1.3b-typescript

Fill In the Middle (FIM)

For FIM tasks, you can use the model to complete code snippets:

function quickSort(arr: number[]): number[] {
  if (arr.length <= 1) {
    return arr;
  }
  const pivot = arr[0];
  const left = [];
  const right = [];