BEE-spoke-data/beecoder-220M-python

1---2license: apache-2.03base_model: BEE-spoke-data/smol_llama-220M-GQA4datasets:5- BEE-spoke-data/pypi_clean-deduped6- bigcode/the-stack-smol-xl7- EleutherAI/proof-pile-28language:9- en10tags:11- python12- codegen13- markdown14- smol_llama15metrics:16- accuracy17inference:18  parameters:19    max_new_tokens: 6420    min_new_tokens: 821    do_sample: true22    epsilon_cutoff: 0.000823    temperature: 0.324    top_p: 0.925    repetition_penalty: 1.0226    no_repeat_ngram_size: 827    renormalize_logits: true28widget:29- text: |30    def add_numbers(a, b):31        return32  example_title: Add Numbers Function33- text: |34    class Car:35        def __init__(self, make, model):36            self.make = make37            self.model = model38 39        def display_car(self):40  example_title: Car Class41- text: |42    import pandas as pd43    data = {'Name': ['Tom', 'Nick', 'John'], 'Age': [20, 21, 19]}44    df = pd.DataFrame(data).convert_dtypes()45    # eda46  example_title: Pandas DataFrame47- text: |48    def factorial(n):49        if n == 0:50            return 151        else:52  example_title: Factorial Function53- text: |54    def fibonacci(n):55        if n <= 0:56            raise ValueError("Incorrect input")57        elif n == 1:58            return 059        elif n == 2:60            return 161        else:62  example_title: Fibonacci Function63- text: |64    import matplotlib.pyplot as plt65    import numpy as np66    x = np.linspace(0, 10, 100)67    # simple plot68  example_title: Matplotlib Plot69- text: |70    def reverse_string(s:str) -> str:71        return72  example_title: Reverse String Function73- text: |74    def is_palindrome(word:str) -> bool:75        return76  example_title: Palindrome Function77- text: |78    def bubble_sort(lst: list):79        n = len(lst)80        for i in range(n):81            for j in range(0, n-i-1):82  example_title: Bubble Sort Function83- text: |84    def binary_search(arr, low, high, x):85        if high >= low:86            mid = (high + low) // 287            if arr[mid] == x:88                return mid89            elif arr[mid] > x:90  example_title: Binary Search Function91pipeline_tag: text-generation92---93 94# BEE-spoke-data/beecoder-220M-python95 96 97 98 99This is `BEE-spoke-data/smol_llama-220M-GQA` fine-tuned for code generation on:100 101- filtered version of stack-smol-XL102- deduped version of 'algebraic stack' from proof-pile-2103- cleaned and deduped pypi (last dataset)104 105This model (and the base model) were both trained using ctx length 2048.  106 107## examples108 109> Example script for inference testing: [here](https://gist.github.com/pszemraj/c7738f664a64b935a558974d23a7aa8c)110 111It has its limitations at 220M, but seems decent for single-line or docstring generation, and/or being used for speculative decoding for such purposes.112 113 114 115![image/png](https://cdn-uploads.huggingface.co/production/uploads/60bccec062080d33f875cd0c/bLrtpr7Vi_MPvtF7mozDN.png)116 117The screenshot is on CPU on a laptop.118 119---