codealpaca
llm-humaneval-benchmarks
codealpaca | llm-humaneval-benchmarks | |
---|---|---|
20 | 10 | |
1,373 | 83 | |
- | - | |
4.4 | 4.9 | |
12 months ago | 11 months ago | |
Python | Jupyter Notebook | |
Apache License 2.0 | MIT License |
Stars - the number of stars that a project has on GitHub. Growth - month over month growth in stars.
Activity is a relative number indicating how actively a project is being developed. Recent commits have higher weight than older ones.
For example, an activity of 9.0 indicates that a project is amongst the top 10% of the most actively developed projects that we are tracking.
codealpaca
-
Just put together a programming performance ranking for popular LLaMAs using the HumanEval+ Benchmark!
CodeAlpaca 7B
-
OpenAI isn’t doing enough to make ChatGPT’s limitations clear
This is great!
Addressing the model limitations a bit: in the demonstration data that is provided to the base model, we should prevent computed or "looked up" answers.
I've seen some of the demonstration data that people are using to train instruction-tuned models and are being taught to respond by making up answers to solutions it shouldn't try to compute. Btw, the output is wrong.
{ "instruction": "What would be the output of the following JavaScript snippet?", "input": "let area = 6 * 5;\nlet radius = area / 3.14;", "output": "The output of the JavaScript snippet is the radius, which is 1.91." }, [1]
The UI note for now would get us very far but by filtering out demonstrations that retrieve or compute information should be filtered out.
Symbol tuning [2] is addressing the quality of demonstrations but we can take it further by removing retrievals and computations altogether.
Bonus: we can demonstrate how to make it respond so that the user/agent be informed of how to compute or retrieve.
1: https://github.com/sahil280114/codealpaca/commit/0d265112c70...
2: https://arxiv.org/abs/2305.08298
- How to Finetune GPT Like Large Language Models on a Custom Dataset
- Ask HN: Those with success using GPT-4 for programming – what are you doing?
-
Is there a colab or guide for fine tuning a 13b model for instruction following?
I found guides like this: https://github.com/sahil280114/codealpaca
-
Can LLMs do static code analysis?
Try, https://github.com/sahil280114/codealpaca, or we’re you trying to stick with more generalist models?
-
LoRA in LLaMAc++? Converting to 4bit? How to use models that are split into multiple .bin ?
Oh, I see. That makes sense. I'm also sleep deprived over here so my reading comprehension is a bit low ;|. Well in that case check out this link: https://github.com/sahil280114/codealpaca
-
Cerebras-GPT: A Family of Open, Compute-Efficient, Large Language Models
Sorry for the late reply, as I said Flan-UL2 (or Flan-T5 if you want lighter models) fine-tuned against a dataset like CodeAlpaca's[0] is probably the best solution if it's intended for commercial use (otherwise LLaMa should perform better).
[0]: https://github.com/sahil280114/codealpaca
- CodeAlpaca – Instruction following code generation model
llm-humaneval-benchmarks
-
LLaMA2 Chat 70B outperformed ChatGPT
You will want to look at HumanEval (https://github.com/abacaj/code-eval) and Eval+ (https://github.com/my-other-github-account/llm-humaneval-ben...) results for coding.
While Llama2 is an improvement over LLaMA v1, it's still nowhere near even the best open models (currently, sans test contamination, WizardCoder-15B, a StarCoder fine tune is at top). It's really not a competition atm though, ChatGPT-4 wipes the floor for coding atm.
-
Claude 2
Since I've been on a AI code-helper kick recently. According to the post, Claude 2 now 71.2%, a significant upgrade from 1.3 (56.0%). It isn't specified whether this is pass@1 or pass@10.
For comparison:
* GPT-4 claims 85.4 on HumanEval, in a recent paper https://arxiv.org/pdf/2303.11366.pdf GPT-4 was tested at 80.1 pass@1 and 91 pass@1 using their Reflexion technique. They also include MBPP and Leetcode Hard benchmark comparisons
* WizardCoder, a StarCoder fine-tune is one of the top open models, scoring a 57.3 pass@1, model card here: https://huggingface.co/WizardLM/WizardCoder-15B-V1.0
* The best open model I know of atm is replit-code-instruct-glaive, a replit-code-3b fine tune, which scores a 63.5% pass@1. An independent developer abacaj has reproduced that announcement as part of code-eval, a repo for getting human-eval results: https://github.com/abacaj/code-eval
Those interested in this area may also want to take a look at this repo https://github.com/my-other-github-account/llm-humaneval-ben... that also ranks with Eval+, the CanAiCode Leaderboard https://huggingface.co/spaces/mike-ravkine/can-ai-code-resul... and airate https://github.com/catid/supercharger/tree/main/airate
Also, as with all LLM evals, to be taken with a grain of salt...
Liu, Jiawei, Chunqiu Steven Xia, Yuyao Wang, and Lingming Zhang. “Is Your Code Generated by ChatGPT Really Correct? Rigorous Evaluation of Large Language Models for Code Generation.” arXiv, June 12, 2023. https://doi.org/10.48550/arXiv.2305.01210.
-
Which LLM works for taboo questions or programming like webscraping?
To get an idea of programming performance, my can-ai-code Leaderboard is freshly updated this morning, but also check out the excellent llm-eval and code-eval leaderboards.
-
Official WizardCoder-15B-V1.0 Released! Can Achieve 59.8% Pass@1 on HumanEval!
❗Note: In this study, we copy the scores for HumanEval and HumanEval+ from the LLM-Humaneval-Benchmarks. Notably, all the mentioned models generate code solutions for each problem utilizing a single attempt, and the resulting pass rate percentage is reported. Our WizardCoder generates answers using greedy decoding and tests with the same code.
-
Hi folks, back with an update to the HumanEval+ programming ranking I posted the other day incorporating your feedback - and some closed models for comparison! Now has improved generation params, new models: Falcon, Starcoder, Codegen, Claude+, Bard, OpenAssistant and more
I switched to RunPod from SageMaker in the middle of this process and boy am I happy I did. It is way cheaper and easier to scale for a project like this, and I highly recommend it. I have a set of tooling to run tests on it en masse now I am happy with - I will try to get my work up on the Github soon!: https://github.com/my-other-github-account/llm-humaneval-benchmarks
- All Model Leaderboards (that I know)
-
Just put together a programming performance ranking for popular LLaMAs using the HumanEval+ Benchmark!
Also, my code I used for this eval is up at https://github.com/my-other-github-account/llm-humaneval-benchmarks/tree/8f3a77eb3508f33a88699aac1c4b10d5e3dc7de1
What are some alternatives?
alpaca.cpp - Locally run an Instruction-Tuned Chat-Style LLM
can-ai-code - Self-evaluating interview for AI coders
alpaca-electron - The simplest way to run Alpaca (and other LLaMA-based local LLMs) on your own computer
code-eval - Run evaluation on LLMs using human-eval benchmark
llm-code - An OpenAI LLM based CLI coding assistant.
WizardLM - Family of instruction-following LLMs powered by Evol-Instruct: WizardLM, WizardCoder and WizardMath
awesome-ai-coding - Awesome AI Coding
text-generation-webui - A Gradio web UI for Large Language Models. Supports transformers, GPTQ, AWQ, EXL2, llama.cpp (GGUF), Llama models.
openplayground-api - A reverse engineered Python API wrapper for OpenPlayground (nat.dev)
llm-leaderboard - A joint community effort to create one central leaderboard for LLMs.
supercharger - Supercharge Open-Source AI Models
ggml - Tensor library for machine learning