metadata
base_model:
- TechxGenus/starcoder2-3b-instruct
tags:
- merge
- mergekit
- lazymergekit
- TechxGenus/starcoder2-3b-instruct
tinyllama-merged-3
tinyllama-merged-3 is a merge of the following models using LazyMergekit:
🧩 Configuration
models:
- model: bigcode/starcoder2-3b
# no parameters necessary for base model
- model: TechxGenus/starcoder2-3b-instruct # follow user intent
parameters:
density:
- filter: mlp.down_proj.4 # specifically targets the 5th layer
value: 0 # assign value of 0 for the 5th layer of down_proj
- value: 1
weight:
- filter: mlp.down_proj
value: [0.3, 0.25, 0.25, 0.15, 0.1]
- filter: mlp.gate_proj
value: [0.7, 0.25, 0.5, 0.45, 0.4]
- filter: mlp.up_proj
value: [0.7, 0.25, 0.5, 0.45, 0.4]
- filter: self_attn
value: [0.7, 0.25, 0.5, 0.45, 0.4]
- value: 1 # fallback for rest of tensors.
tokenizer_source: union
merge_method: dare_ties
base_model: bigcode/starcoder2-3b
parameters:
normalize: true
int8_mask: true
dtype: bfloat16
💻 Usage
!pip install -qU transformers accelerate
from transformers import AutoTokenizer
import transformers
import torch
model = "choprahetarth/tinyllama-merged-3"
messages = [{"role": "user", "content": "What is a large language model?"}]
tokenizer = AutoTokenizer.from_pretrained(model)
prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
pipeline = transformers.pipeline(
"text-generation",
model=model,
torch_dtype=torch.float16,
device_map="auto",
)
outputs = pipeline(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
print(outputs[0]["generated_text"])