File size: 2,187 Bytes
b873af7 b8a65f6 9e2b422 5b66cdc 9e2b422 67ed110 b8a65f6 5b66cdc 9e2b422 b8a65f6 16843a5 b8a65f6 5b66cdc b8a65f6 67ed110 5b66cdc b8a65f6 5b66cdc b8a65f6 67ed110 b8a65f6 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 |
---
license: cc-by-sa-4.0
---
<div style="width: 800px; margin: auto;">
<h2>Model Description</h2>
<p>“Luna AI Llama2 Uncensored” is a Llama2 based Chat model <br />fine-tuned on over 40,000 instructions. <br />
This model was fine-tuned by Tap, the creator of Luna AI. <br />
The result is an enhanced Llama2 7b model that rivals ChatGPT in performance <br />across a variety of tasks.</p>
<p>This model stands out for its long responses, low hallucination rate, and absence of censorship mechanisms. <br />
The fine-tuning process was performed on an 8x a100 80GB machine.</p>
<h2>Model Training</h2>
<p>The model was trained almost entirely on synthetic outputs. <br />
This includes data from diverse sources such as Orca, GPT4-LLM,<br /> and a custom unreleased dataset.<br />
The total volume of data encompassed over 40,000 high quality instructions.</p>
<p>Huge shoutout and acknowledgement is deserved for all the dataset creators who generously share their datasets openly.</p>
<h2>Prompt Format</h2>
<p>The model follows the Vicuna 1.1/ OpenChat format:</p>
<h2>Future Plans</h2>
<p>The model is currently being uploaded in FP16 format, <br />and there are plans to convert the model to GGML and GPTQ 4bit quantizations.</p>
<h2>Benchmark Results</h2>
<pre>
| Task |Version| Metric |Value | |Stderr|
|-------------|------:|--------|-----:|---|-----:|
|TO be announced
</pre>
<h2>Ethical considerations</h2>
<p>The data used to train the model is collected from various sources, mostly from the Web. <br />
As such, it contains offensive, harmful and biased content. <br />We thus expect the model to exhibit such biases from the training data.</p>
<h2>Human life</h2>
<p>The model is not intended to inform decisions about matters central to human life, <br />and should not be used in such a way.</p>
<h2>Risks and harms</h2>
<p>Risks and harms of large language models include the generation of harmful, offensive or biased content. <br />
These models are often prone to generating incorrect information, sometimes referred to as hallucinations.
<br /> We do not expect our model to be an exception in this regard.</p>
</div>
|