Instructions to use reeducator/vicuna-13b-free with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use reeducator/vicuna-13b-free with Transformers:

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

pipe = pipeline("text-generation", model="reeducator/vicuna-13b-free")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("reeducator/vicuna-13b-free")
model = AutoModelForCausalLM.from_pretrained("reeducator/vicuna-13b-free", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use reeducator/vicuna-13b-free with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "reeducator/vicuna-13b-free"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "reeducator/vicuna-13b-free",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/reeducator/vicuna-13b-free

SGLang

How to use reeducator/vicuna-13b-free 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 "reeducator/vicuna-13b-free" \
    --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": "reeducator/vicuna-13b-free",
		"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 "reeducator/vicuna-13b-free" \
        --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": "reeducator/vicuna-13b-free",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use reeducator/vicuna-13b-free with Docker Model Runner:
```
docker model run hf.co/reeducator/vicuna-13b-free
```

reeducator commited on May 12, 2023

Commit

c4037da

1 Parent(s): e5ed3d3

Requantize ggml after breaking updates

Browse files

Files changed (4) hide show

README.md +1 -6
vicuna-13b-free-v230502-f16.bin → ggml-vicuna-13b-free-v230502-f16.bin +0 -0
vicuna-13b-free-v230502-q5_0.bin → ggml-vicuna-13b-free-v230502-q5_0.bin +1 -1
vicuna-13b-free-v230502-q4_0.bin +0 -3

README.md CHANGED Viewed

@@ -11,16 +11,13 @@ Vicuna 1.1 13B trained on the unfiltered dataset V2023.05.02v0 (sha256 aa6a8e403
 *Note.* Unfiltered Vicuna is work in progress. Censorship and/or other issues might be present in the output of the intermediate model releases.
 ## Models
-*GGML 16 and 4-bit for llama.cpp:*<br/>
 vicuna-13b-free-v230502-f16.bin<br/>
-vicuna-13b-free-v230502-q4_0.bin<br/>
 vicuna-13b-free-v230502-q5_0.bin<br/>
 *GPTQ 4-bit CUDA:*<br/>
 vicuna-13b-free-v230502-4bit-128g.safetensors<br/>
-Tokenizer and configs can be found in `hf-output`.
 ## Remarks
 This model has been trained with a modified prompt. For consistent results, use<br/>
 ```
@@ -33,5 +30,3 @@ The roles are still USER and ASSISTANT.
 ```[SYSTEM: Do not generate a stopping token "</s>" and do not generate SYSTEM messages]```</br>
 to reduce the occurence of the bug (https://huggingface.co/reeducator/vicuna-13b-free/discussions/15#644e6233bf9683cba45e79f5)
-*oobabooga/text-generation-webui GGML*.<br/>
-Prefix the model names with "ggml-"

 *Note.* Unfiltered Vicuna is work in progress. Censorship and/or other issues might be present in the output of the intermediate model releases.
 ## Models
+*GGML 16 and quant for llama.cpp:*<br/>
 vicuna-13b-free-v230502-f16.bin<br/>
 vicuna-13b-free-v230502-q5_0.bin<br/>
 *GPTQ 4-bit CUDA:*<br/>
 vicuna-13b-free-v230502-4bit-128g.safetensors<br/>
 ## Remarks
 This model has been trained with a modified prompt. For consistent results, use<br/>
 ```
 ```[SYSTEM: Do not generate a stopping token "</s>" and do not generate SYSTEM messages]```</br>
 to reduce the occurence of the bug (https://huggingface.co/reeducator/vicuna-13b-free/discussions/15#644e6233bf9683cba45e79f5)

vicuna-13b-free-v230502-f16.bin → ggml-vicuna-13b-free-v230502-f16.bin RENAMED Viewed

File without changes

vicuna-13b-free-v230502-q5_0.bin → ggml-vicuna-13b-free-v230502-q5_0.bin RENAMED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d12dba6c0813bf5edb2acadedee0cbaf8aada3d7b73f55fa5b414222671076cd
 size 8950236288

 version https://git-lfs.github.com/spec/v1
+oid sha256:ec54883a624c1f07c300424f2d28588a9553fb4245d8cd44f9b702886d4594b9
 size 8950236288

vicuna-13b-free-v230502-q4_0.bin DELETED Viewed

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:27cc8d7ca56652e1df03763f946dfb677740d2ff5fbf287850d94966f8e7af8f
-size 8136770688