File size: 1,495 Bytes
c911d26 50a405f c911d26 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 |
---
license: apache-2.0
base_model:
- mistralai/Mistral-7B-Instruct-v0.3
datasets:
- samzheng/SymbolicCode-CoT
language:
- en
pipeline_tag: question-answering
---
model_name: "Mistral-7B-Math (Merged FP16 Checkpoint)" \
repo: "samzheng/mistral-7b-math-merged"\
base_model:
name: "unsloth/mistral-7b-instruct-v0.3-bnb-4bit"
url: "https://huggingface.co/unsloth/mistral-7b-instruct-v0.3-bnb-4bit"\
task: "Grade-school symbolic math word problems → Python code answers"\
fine_tuning:
method: "LoRA adapters (r=16, α=16, dropout=0) merged into the base weights, FP16 precision"
parameters:
r: 16
alpha: 16
dropout: 0\
dataset:
description: "6.7k Alpaca-formatted Q/A pairs with chain-of-thought + code"
splits:
- "symboliccode_cot_train"
- "symboliccode_cot_validation"\
language: python \
code:
from transformers import AutoModelForCausalLM, AutoTokenizer
model = AutoModelForCausalLM.from_pretrained(
"samzheng/mistral-7b-math-merged",
torch_dtype="auto", device_map="auto"
)
tok = AutoTokenizer.from_pretrained("samzheng/mistral-7b-math-merged")
prompt = """Below is an instruction that describes a task...
### Instruction: Solve the problem using step-by-step reasoning and provide Python code.
### Input: Solve for x: 2x + 5 = 17
### Response:
"""
print(tok.decode(model.generate(**tok(prompt, return_tensors="pt").to(model.device),max_new_tokens=256)[0], skip_special_tokens=True)) |