Instructions to use TokenBender/llama2-7b-chat-hf-codeCherryPop-qLoRA-merged with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use TokenBender/llama2-7b-chat-hf-codeCherryPop-qLoRA-merged with Transformers:

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

pipe = pipeline("text-generation", model="TokenBender/llama2-7b-chat-hf-codeCherryPop-qLoRA-merged")

# Load model directly
from transformers import AutoTokenizer, AutoModelForMultimodalLM

tokenizer = AutoTokenizer.from_pretrained("TokenBender/llama2-7b-chat-hf-codeCherryPop-qLoRA-merged")
model = AutoModelForMultimodalLM.from_pretrained("TokenBender/llama2-7b-chat-hf-codeCherryPop-qLoRA-merged")

Inference
Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use TokenBender/llama2-7b-chat-hf-codeCherryPop-qLoRA-merged with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "TokenBender/llama2-7b-chat-hf-codeCherryPop-qLoRA-merged"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "TokenBender/llama2-7b-chat-hf-codeCherryPop-qLoRA-merged",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/TokenBender/llama2-7b-chat-hf-codeCherryPop-qLoRA-merged

SGLang

How to use TokenBender/llama2-7b-chat-hf-codeCherryPop-qLoRA-merged 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 "TokenBender/llama2-7b-chat-hf-codeCherryPop-qLoRA-merged" \
    --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": "TokenBender/llama2-7b-chat-hf-codeCherryPop-qLoRA-merged",
		"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 "TokenBender/llama2-7b-chat-hf-codeCherryPop-qLoRA-merged" \
        --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": "TokenBender/llama2-7b-chat-hf-codeCherryPop-qLoRA-merged",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use TokenBender/llama2-7b-chat-hf-codeCherryPop-qLoRA-merged with Docker Model Runner:
```
docker model run hf.co/TokenBender/llama2-7b-chat-hf-codeCherryPop-qLoRA-merged
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

YAML Metadata Warning:empty or missing yaml metadata in repo card

Check out the documentation for more information.

Overview:

description:

This is a llama2 7B HF chat model fine-tuned on 122k code instructions. In my early experiments it seems to be doing very well.

additional_info:

It's a bottom of the barrel model 😂 but after quantization it can be
valuable for sure. It definitely proves that a 7B can be useful for boilerplate
code stuff though.

Plans:

next_steps: "I've a few things in mind and after that this will be more valuable."

tasks:

- name: "I'll quantize these"
  timeline: "Possibly tonight or tomorrow in the day"
  result: "Then it can be run locally with 4G ram."
- name: "I've used alpaca style instruction tuning"
  improvement: |
    I'll switch to llama2 style [INST]<<SYS>> style and see if
    it improves anything.
- name: "HumanEval report and checking for any training data leaks"
- attempt: "I'll try 8k context via RoPE enhancement"
  hypothesis: "Let's see if that degrades performance or not."

commercial_use: | So far I think this can be used commercially but this is a adapter on Meta's llama2 with some gating issues so that is there. contact_info: "If you find any issues or want to just holler at me, you can reach out to me - https://twitter.com/4evaBehindSOTA"

Library:

name: "peft"

Training procedure:

quantization_config: load_in_8bit: False load_in_4bit: True llm_int8_threshold: 6.0 llm_int8_skip_modules: None llm_int8_enable_fp32_cpu_offload: False llm_int8_has_fp16_weight: False bnb_4bit_quant_type: "nf4" bnb_4bit_use_double_quant: False bnb_4bit_compute_dtype: "float16"

Framework versions:

PEFT: "0.5.0.dev0"

Downloads last month: 6

Model tree for TokenBender/llama2-7b-chat-hf-codeCherryPop-qLoRA-merged

Finetunes

2 models

Quantizations

3 models

Collection including TokenBender/llama2-7b-chat-hf-codeCherryPop-qLoRA-merged

PIC (Partner-in-Crime) project

Collection

Empathetic, small, really useful personalised models. • 3 items • Updated Dec 10, 2023 • 2