Instructions to use brucethemoose/Capybara-Tess-Yi-34B-200K-DARE-Ties with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use brucethemoose/Capybara-Tess-Yi-34B-200K-DARE-Ties with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="brucethemoose/Capybara-Tess-Yi-34B-200K-DARE-Ties")

# Load model directly
from transformers import AutoTokenizer, AutoModelForMultimodalLM

tokenizer = AutoTokenizer.from_pretrained("brucethemoose/Capybara-Tess-Yi-34B-200K-DARE-Ties")
model = AutoModelForMultimodalLM.from_pretrained("brucethemoose/Capybara-Tess-Yi-34B-200K-DARE-Ties")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use brucethemoose/Capybara-Tess-Yi-34B-200K-DARE-Ties with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "brucethemoose/Capybara-Tess-Yi-34B-200K-DARE-Ties"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "brucethemoose/Capybara-Tess-Yi-34B-200K-DARE-Ties",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/brucethemoose/Capybara-Tess-Yi-34B-200K-DARE-Ties

SGLang

How to use brucethemoose/Capybara-Tess-Yi-34B-200K-DARE-Ties with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "brucethemoose/Capybara-Tess-Yi-34B-200K-DARE-Ties" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "brucethemoose/Capybara-Tess-Yi-34B-200K-DARE-Ties",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "brucethemoose/Capybara-Tess-Yi-34B-200K-DARE-Ties" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "brucethemoose/Capybara-Tess-Yi-34B-200K-DARE-Ties",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use brucethemoose/Capybara-Tess-Yi-34B-200K-DARE-Ties with Docker Model Runner:
```
docker model run hf.co/brucethemoose/Capybara-Tess-Yi-34B-200K-DARE-Ties
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

CaPlatTessDolXaBoros-Yi-34B-200K-DARE-Ties

NousResearch/Nous-Capybara-34B, migtissera/Tess-M-v1.2 and migtissera/Tess-M-v1.3 merged with a new, experimental implementation of "dare ties" via mergekit. See:

Language Models are Super Mario: Absorbing Abilities from Homologous Models as a Free Lunch

https://github.com/yule-BUAA/MergeLM

https://github.com/cg123/mergekit/tree/dare-tokenizer

Highly experimental and still being tested! But this should yield a better merge than a typical linear/slerp merge or even a ties merge.

Merged with the following config, and the tokenizer from Yi Llamafied:

models:
  - model: /home/alpha/Storage/Models/Raw/larryvrh_Yi-34B-200K-Llamafied
    # no parameters necessary for base model
  - model: /home/alpha/Storage/Models/Raw/migtissera_Tess-M-v1.3
    parameters:
      weight: 0.50
      density: 0.56
  - model: /home/alpha/Storage/Models/Raw/migtissera_Tess-M-v1.2
    parameters:
      weight: 0.20
      density: 0.50
  - model: /home/alpha/Storage/Models/Raw/Nous-Capybara-34B
    parameters:
      weight: 0.50
      density: 0.56
merge_method: dare_ties
base_model: /home/alpha/Storage/Models/Raw/larryvrh_Yi-34B-200K-Llamafied
parameters:
  int8_mask: true
dtype: bfloat16

Tess 1.2 (at a low weight) and 1.3 were used because, according to the trainer, they were trained on different datasets: https://migel.substack.com/p/learnings-from-training-tess

As the Tess creator warned about, if the model repeats at high context like Tess 1.2, let me know!

I chose not to include other finetunes, such as Dolphin, because they aren't trained on the 200K base. If any other 200K finetunes pop up, let me know.

Prompt template: Orca-Vicuna

SYSTEM: {system_message}
USER: {prompt}
ASSISTANT:

Being a Yi model, try disabling the BOS token and/or running a lower temperature with MinP if output doesn't seem right.

Sometimes the model "spells out" the stop token as </s> like Capybara, so you may need to add </s> as an additional stopping condition.

Credits:

https://github.com/cg123/mergekit/tree/dare-tokenizer

https://huggingface.co/NousResearch/Nous-Capybara-34B/

https://huggingface.co/migtissera/Tess-M-v1.2

https://huggingface.co/migtissera/Tess-M-v1.3

https://huggingface.co/larryvrh/Yi-34B-200K-Llamafied

https://huggingface.co/01-ai/Yi-34B-200K

Downloads last month: 7

Safetensors

Model size

34B params

Tensor type

BF16

Model tree for brucethemoose/Capybara-Tess-Yi-34B-200K-DARE-Ties

Quantizations

5 models