README.md

---
inference: false
license: other
---

<!-- header start -->
<div style="width: 100%;">
    <img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;">
</div>
<div style="display: flex; justify-content: space-between; width: 100%;">
    <div style="display: flex; flex-direction: column; align-items: flex-start;">
        <p><a href="https://discord.gg/Jq4vkcDakD">Chat & support: my new Discord server</a></p>
    </div>
    <div style="display: flex; flex-direction: column; align-items: flex-end;">
        <p><a href="https://www.patreon.com/TheBlokeAI">Want to contribute? TheBloke's Patreon page</a></p>
    </div>
</div>
<!-- header end -->

# OpenAccess AI Collective's Manticore 13B GGML

These files are GGML format model files for [OpenAccess AI Collective's Manticore 13B](https://huggingface.co/openaccess-ai-collective/manticore-13b).

GGML files are for CPU + GPU inference using [llama.cpp](https://github.com/ggerganov/llama.cpp) and libraries and UIs which support this format, such as:
* [text-generation-webui](https://github.com/oobabooga/text-generation-webui)
* [KoboldCpp](https://github.com/LostRuins/koboldcpp)
* [ParisNeo/GPT4All-UI](https://github.com/ParisNeo/gpt4all-ui)
* [llama-cpp-python](https://github.com/abetlen/llama-cpp-python)
* [ctransformers](https://github.com/marella/ctransformers)

## Repositories available

* [4-bit GPTQ models for GPU inference](https://huggingface.co/TheBloke/Manticore-13B-GPTQ)
* [2, 3, 4, 5, 6 and 8-bit GGML models for CPU+GPU inference](https://huggingface.co/TheBloke/Manticore-13B-GGML)
* [Unquantised fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/openaccess-ai-collective/manticore-13b)

<!-- compatibility_ggml start -->
## Compatibility

### Original llama.cpp quant methods: `q4_0, q4_1, q5_0, q5_1, q8_0`

I have quantized these 'original' quantisation methods using an older version of llama.cpp so that they remain compatible with llama.cpp as of May 19th, commit `2d5db48`.

They should be compatible with all current UIs and libraries that use llama.cpp, such as those listed at the top of this README.

### New k-quant methods: `q2_K, q3_K_S, q3_K_M, q3_K_L, q4_K_S, q4_K_M, q5_K_S, q6_K`

These new quantisation methods are only compatible with llama.cpp as of June 6th, commit `2d43387`.

They will NOT be compatible with koboldcpp, text-generation-ui, and other UIs and libraries yet. Support is expected to come over the next few days.

## Explanation of the new k-quant methods

The new methods available are:
* GGML_TYPE_Q2_K - "type-1" 2-bit quantization in super-blocks containing 16 blocks, each block having 16 weight. Block scales and mins are quantized with 4 bits. This ends up effectively using 2.5625 bits per weight (bpw)
* GGML_TYPE_Q3_K - "type-0" 3-bit quantization in super-blocks containing 16 blocks, each block having 16 weights. Scales are quantized with 6 bits. This end up using 3.4375 bpw.
* GGML_TYPE_Q4_K - "type-1" 4-bit quantization in super-blocks containing 8 blocks, each block having 32 weights. Scales and mins are quantized with 6 bits. This ends up using 4.5 bpw.
* GGML_TYPE_Q5_K - "type-1" 5-bit quantization. Same super-block structure as GGML_TYPE_Q4_K resulting in 5.5 bpw
* GGML_TYPE_Q6_K - "type-0" 6-bit quantization. Super-blocks with 16 blocks, each block having 16 weights. Scales are quantized with 8 bits. This ends up using 6.5625 bpw
* GGML_TYPE_Q8_K - "type-0" 8-bit quantization. Only used for quantizing intermediate results. The difference to the existing Q8_0 is that the block size is 256. All 2-6 bit dot products are implemented for this quantization type.

Refer to the Provided Files table below to see what files use which methods, and how.
<!-- compatibility_ggml end -->

## Provided files
| Name | Quant method | Bits | Size | Max RAM required | Use case |
| ---- | ---- | ---- | ---- | ---- | ----- |
| Manticore-13B.ggmlv3.q2_K.bin | q2_K | 2 | 5.43 GB | 7.93 GB | New k-quant method. Uses GGML_TYPE_Q4_K for the attention.vw and feed_forward.w2 tensors, GGML_TYPE_Q2_K for the other tensors. |
| Manticore-13B.ggmlv3.q3_K_L.bin | q3_K_L | 3 | 6.87 GB | 9.37 GB | New k-quant method. Uses GGML_TYPE_Q5_K for the attention.wv, attention.wo, and feed_forward.w2 tensors, else GGML_TYPE_Q3_K |
| Manticore-13B.ggmlv3.q3_K_M.bin | q3_K_M | 3 | 6.25 GB | 8.75 GB | New k-quant method. Uses GGML_TYPE_Q4_K for the attention.wv, attention.wo, and feed_forward.w2 tensors, else GGML_TYPE_Q3_K |
| Manticore-13B.ggmlv3.q3_K_S.bin | q3_K_S | 3 | 5.59 GB | 8.09 GB | New k-quant method. Uses GGML_TYPE_Q3_K for all tensors |
| Manticore-13B.ggmlv3.q4_0.bin | q4_0 | 4 | 7.32 GB | 9.82 GB | Original llama.cpp quant method, 4-bit. |
| Manticore-13B.ggmlv3.q4_1.bin | q4_1 | 4 | 8.14 GB | 10.64 GB | Original llama.cpp quant method, 4-bit. Higher accuracy than q4_0 but not as high as q5_0. However has quicker inference than q5 models. |
| Manticore-13B.ggmlv3.q4_K_M.bin | q4_K_M | 4 | 7.82 GB | 10.32 GB | New k-quant method. Uses GGML_TYPE_Q6_K for half of the attention.wv and feed_forward.w2 tensors, else GGML_TYPE_Q4_K |
| Manticore-13B.ggmlv3.q4_K_S.bin | q4_K_S | 4 | 7.32 GB | 9.82 GB | New k-quant method. Uses GGML_TYPE_Q4_K for all tensors |
| Manticore-13B.ggmlv3.q5_0.bin | q5_0 | 5 | 8.95 GB | 11.45 GB | Original llama.cpp quant method, 5-bit. Higher accuracy, higher resource usage and slower inference. |
| Manticore-13B.ggmlv3.q5_1.bin | q5_1 | 5 | 9.76 GB | 12.26 GB | Original llama.cpp quant method, 5-bit. Even higher accuracy, resource usage and slower inference. |
| Manticore-13B.ggmlv3.q5_K_M.bin | q5_K_M | 5 | 9.21 GB | 11.71 GB | New k-quant method. Uses GGML_TYPE_Q6_K for half of the attention.wv and feed_forward.w2 tensors, else GGML_TYPE_Q5_K |
| Manticore-13B.ggmlv3.q5_K_S.bin | q5_K_S | 5 | 8.95 GB | 11.45 GB | New k-quant method. Uses GGML_TYPE_Q5_K for all tensors |
| Manticore-13B.ggmlv3.q6_K.bin | q6_K | 6 | 10.68 GB | 13.18 GB | New k-quant method. Uses GGML_TYPE_Q8_K - 6-bit quantization - for all tensors |
| Manticore-13B.ggmlv3.q8_0.bin | q8_0 | 8 | 13.83 GB | 16.33 GB | Original llama.cpp quant method, 8-bit. Almost indistinguishable from float16. High resource use and slow. Not recommended for most users. |


**Note**: the above RAM figures assume no GPU offloading. If layers are offloaded to the GPU, this will reduce RAM usage and use VRAM instead.

## How to run in `llama.cpp`

I use the following command line; adjust for your tastes and needs:

```
./main -t 10 -ngl 32 -m Manticore-13B.ggmlv3.q5_0.bin --color -c 2048 --temp 0.7 --repeat_penalty 1.1 -n -1 -p "### Instruction: Write a story about llamas\n### Response:"
```
Change `-t 10` to the number of physical CPU cores you have. For example if your system has 8 cores/16 threads, use `-t 8`.

Change `-ngl 32` to the number of layers to offload to GPU. Remove it if you don't have GPU acceleration.

If you want to have a chat-style conversation, replace the `-p <PROMPT>` argument with `-i -ins`

## How to run in `text-generation-webui`

Further instructions here: [text-generation-webui/docs/llama.cpp-models.md](https://github.com/oobabooga/text-generation-webui/blob/main/docs/llama.cpp-models.md).

<!-- footer start -->
## Discord

For further support, and discussions on these models and AI in general, join us at:

[TheBloke AI's Discord server](https://discord.gg/Jq4vkcDakD)

## Thanks, and how to contribute.

Thanks to the [chirper.ai](https://chirper.ai) team!

I've had a lot of people ask if they can contribute. I enjoy providing models and helping people, and would love to be able to spend even more time doing it, as well as expanding into new projects like fine tuning/training.

If you're able and willing to contribute it will be most gratefully received and will help me to keep providing more models, and to start work on new AI projects.

Donaters will get priority support on any and all AI/LLM/model questions and requests, access to a private Discord room, plus other benefits.

* Patreon: https://patreon.com/TheBlokeAI
* Ko-Fi: https://ko-fi.com/TheBlokeAI

**Special thanks to**: Luke from CarbonQuill, Aemon Algiz, Dmitriy Samsonov.

**Patreon special mentions**: Ajan Kanaga, Kalila, Derek Yates, Sean Connelly, Luke, Nathan LeClaire, Trenton Dambrowitz, Mano Prime, David Flickinger, vamX, Nikolai Manek, senxiiz, Khalefa Al-Ahmad, Illia Dulskyi, trip7s trip, Jonathan Leane, Talal Aujan, Artur Olbinski, Cory Kujawski, Joseph William Delisle, Pyrater, Oscar Rangel, Lone Striker, Luke Pendergrass, Eugene Pentland, Johann-Peter Hartmann.

Thank you to all my generous patrons and donaters!

<!-- footer end -->

# Original model card: OpenAccess AI Collective's Manticore 13B


# Manticore 13B - (previously Wizard Mega)

**[💵 Donate to OpenAccess AI Collective](https://github.com/sponsors/OpenAccess-AI-Collective) to help us keep building great tools and models!**

![open-llm-leaderboard.png](https://huggingface.co/openaccess-ai-collective/manticore-13b/resolve/main/open-llm-leaderboard.png)

Questions, comments, feedback, looking to donate, or want to help? Reach out on our [Discord](https://discord.gg/EqrvvehG) or email [[email protected]](mailto:[email protected])

Manticore 13B is a Llama 13B model fine-tuned on the following datasets:
- [ShareGPT](https://huggingface.co/datasets/anon8231489123/ShareGPT_Vicuna_unfiltered) - based on a cleaned and de-suped subset
- [WizardLM](https://huggingface.co/datasets/ehartford/WizardLM_alpaca_evol_instruct_70k_unfiltered)
- [Wizard-Vicuna](https://huggingface.co/datasets/ehartford/wizard_vicuna_70k_unfiltered)
- [subset of QingyiSi/Alpaca-CoT for roleplay and CoT](https://huggingface.co/QingyiSi/Alpaca-CoT)
- [GPT4-LLM-Cleaned](https://huggingface.co/datasets/teknium/GPT4-LLM-Cleaned)
- [GPTeacher-General-Instruct](https://huggingface.co/datasets/teknium/GPTeacher-General-Instruct)
- ARC-Easy & ARC-Challenge - instruct augmented  for detailed responses
- mmlu: instruct augmented for detailed responses subset including
  - abstract_algebra
  - conceptual_physics
  - formal_logic
  - high_school_physics
  - logical_fallacies
- [hellaswag](https://huggingface.co/datasets/hellaswag) - 5K row subset of instruct augmented for concise responses
- [metaeval/ScienceQA_text_only](https://huggingface.co/datasets/metaeval/ScienceQA_text_only) - instruct for concise responses
- [openai/summarize_from_feedback](https://huggingface.co/datasets/openai/summarize_from_feedback) - instruct augmented tl;dr summarization


# Demo

Try out the model in HF Spaces. The demo uses a quantized GGML version of the model to quickly return predictions on smaller GPUs (and even CPUs). Quantized GGML may have some minimal loss of model quality.
- https://huggingface.co/spaces/openaccess-ai-collective/manticore-ggml

## Release Notes

- https://wandb.ai/wing-lian/manticore-13b/runs/nq3u3uoh/workspace

## Build

Manticore was built with [Axolotl](https://github.com/OpenAccess-AI-Collective/axolotl) on 8xA100 80GB 
 - Preview Release: 3 epochs taking approximately 24 hours. 
 - The configuration to duplicate this build is provided in this repo's [/config folder](https://huggingface.co/openaccess-ai-collective/manticore-13b/tree/main/configs).

## Bias, Risks, and Limitations
Manticore has not been aligned to human preferences with techniques like RLHF or deployed with in-the-loop filtering of responses like ChatGPT, so the model can produce problematic outputs (especially when prompted to do so).
Manticore was fine-tuned from the base model LlaMa 13B, please refer to its model card's Limitations Section for relevant information.

## Examples

````
### Instruction: write Python code that returns the first n numbers of the Fibonacci sequence using memoization.

### Assistant:
Here’s an example of how you can use memoization to efficiently calculate the first n numbers of the Fibonacci sequence in Python:
```
def fib(n):
 # Memoize the last calculated Fibonacci number for each index
 cache = {}
 
 def fib_memoized(n):
 if n not in cache:
 cache[n] = [0, 1]
 else:
 return cache[n][-1], cache[n][-2]
 
 # Recursive helper function to calculate the Fibonacci sequence up to a given index
 def fib_helper(n):
 if n == 0 or n == 1:
 return 0, 1
 else:
 return fib_helper(n - 1) + fib_helper(n - 2)
 
# Calculate the first n numbers of the Fibonacci sequence using memoization
 def fib_memoized_seq(n):
 if n == 0 or n == 1:
 return 0, 1
 else:
 result = fib_helper(n - 1) + fib_helper(n - 2)
 cache[n] = [result]
 return result, cache[n][-1], cache[n][-2]
 
 # Example usage
 for i in range(len(fib_memoized_seq(3))):
 print(fib_memoized_seq(3)[i])
```

In this implementation, we use a cache dictionary to memoize the last calculated Fibonacci number for each index. We define two helper functions: fib_helper, which recursively calculates the Fibonacci sequence up to a given index, and fib_memoized, which uses memoization to efficiently calculate the first n numbers of the Fibonacci sequence. Finally, we define the fib_memoized_seq function, which is our main function for calculating the first n numbers of the Fibonacci sequence using memoization.
````

```
USER: Finish the joke, a mechanic and a car salesman walk into a bar...  
ASSISTANT: … and when they get up to leave, they realize that they’ve left their keys in their cars. The mechanic says “I know how to fix this” and opens his toolbox to grab something. Then he stops and looks at the salesman. He says “You drive a Toyota, right?” The salesman nods. The mechanic then takes out a tiny screwdriver from his toolbox and starts working on the lock of the car door. The salesman is amazed and asks how he knew that. The mechanic replies “I can tell by the way you parked your car.”
```

```
USER: Describe how a tsunami would affect the residents of Denver Colorado
ASSISTANT: The impact of a tsunami in Denver, Colorado would likely be minimal. As there are no coastal areas or bodies of water nearby, a tsunami generated from an earthquake or volcanic activity elsewhere in the world would have little effect on the city and its residents. Even if a large tsunami were to hit the coastline of California or Mexico, the distance between Denver and these locations means that any impacts on the city itself would be minimal. 
```

```
USER: If you could do one thing right now, what would it be?
ASSISTANT: I’d hug my loved ones a little tighter today and tell them how much they mean to me. 
USER: And who are your loved ones?
ASSISTANT: My closest friends and family members.
USER: What are their names?
ASSISTANT: My two best friends are named Emily and Jessica, my mom is Mary, and my dad is David. Answering the question with a bit more information about who they are or what they mean to me would help provide a more personalized response.
```