Mistral-7B-Chemical-NER

Model Card for Model ID

<!-- Provide a quick summary of what the model is/does. -->

A LoRA-adapted version of Mistral-7B-Instruct-v0.1 trained to act as a lightweight chemical/biomedical judge (scoring or classifying completions given a prompt) using a simple prompt → completion JSONL dataset. The model is intended to evaluate or label biomedical-related text outputs rather than generate long-form content.

Model Details

Model Description

<!-- Provide a longer summary of what this model is. -->

This model fine-tunes the base Mistral-7B-Instruct-v0.1 with LoRA adapters (PEFT) on a JSONL dataset of pairs: prompt: the instruction/context to judge completion: the desired judged output/label/explanation Only the completion segment contributes to the loss (the prompt tokens are masked with -100), matching your training script’s supervision scheme. The result is a parameter-efficient adapter suitable for inference on commodity GPUs with 8-bit base weights.

• Developed by: Idrissa M. Dicko
• Funded by [optional]: [More Information Needed]
• Shared by [optional]: [More Information Needed]
• Model type: [More Information Needed]
• Language(s) (NLP): [More Information Needed]
• License: [More Information Needed]
• Finetuned from model [optional]: /Mistral-7B-Instruct-v0.1

Model Sources [optional]

<!-- Provide the basic links for the model. -->

• Repository: https://github.com/idrissa-gif/EXTRACTION_DATA_DRUG_CHEMICALS
• Paper [optional]: [More Information Needed]
• Demo [optional]: [More Information Needed]

Uses

<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->

Direct Use

<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->

As a judge/evaluator for chemical-related text outputs: given a prompt (task/context) and a candidate answer, the model produces a concise judgment/label/score rationale depending on how you format your prompt.

Downstream Use [optional]

<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->

Incorporate the LoRA adapter in a larger pipeline that generates candidates and then ranks/filters them using this judge.

Out-of-Scope Use

<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->

• Medical advice, diagnosis, treatment recommendations.
• Designing new chemicals/drugs or suggesting synthesis routes.
• Safety-critical decisions without human oversight.
• Any use that conflicts with the base model’s license or content policy.

Bias, Risks, and Limitations

<!-- This section is meant to convey both technical and sociotechnical limitations. -->

Hallucinations: As a generative LM, it may produce confident but incorrect judgments. Domain bias: Behavior reflects the distribution and labeling of your training data. Ambiguity sensitivity: Judgments may vary with small prompt changes; provide clear, structured prompts. Security / Safety: Do not rely on the model for hazardous chemical substance guidance.

Recommendations

<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->

• Keep a human-in-the-loop for any consequential evaluation.
• Use structured prompts (templates) and, where possible, few-shot exemplars to stabilize outputs.
• Consider calibrating outputs (e.g., convert free-text judgments to discrete labels).

How to Get Started with the Model

Use the code below to get started with the model.

from transformers import AutoTokenizer, AutoModelForCausalLM
from peft import PeftModel
import torch

base = "mistralai/Mistral-7B-Instruct-v0.1"
adapter_dir = "judge_llama_lora"  # your trained output dir

tok = AutoTokenizer.from_pretrained(base, use_fast=False)
if tok.pad_token is None:
    tok.pad_token = tok.eos_token

base_model = AutoModelForCausalLM.from_pretrained(
    base,
    load_in_8bit=True,          # matches training recipe
    device_map="auto",
)
model = PeftModel.from_pretrained(base_model, adapter_dir)
model.config.pad_token_id = tok.eos_token_id

prompt = "### Instruction:\nEvaluate the following chemical statement...\n### Candidate:\n..."
inputs = tok(prompt, return_tensors="pt").to(model.device)
with torch.inference_mode():
    out = model.generate(**inputs, max_new_tokens=128)
print(tok.decode(out[0], skip_special_tokens=True))

Training Details

Training Data

<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->

• BioRED (Biomedical Research Entity and Relation Dataset)
• BC5CHEMD (BioCreative V Chemical Disease Corpus for chemicals)
• BC4CDR (BioCreative IV Chemical-Disease Relations corpus)

Training Procedure

<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->

Preprocessing [optional]

Tokenized with AutoTokenizer.from_pretrained(base_model, use_fast=False) pad_token set to eos_token Fixed max_length = 128 with truncation/padding to max length.

Training Hyperparameters

• Training regime: [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->

Speeds, Sizes, Times [optional]

<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->

[More Information Needed]

Evaluation

<!-- This section describes the evaluation protocols and provides the results. -->

Testing Data, Factors & Metrics

Testing Data

<!-- This should link to a Dataset Card if possible. -->

[More Information Needed]

Factors

<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->

[More Information Needed]

Metrics

<!-- These are the evaluation metrics being used, ideally with a description of why. -->

[More Information Needed]

Results

[More Information Needed]

Summary

Model Examination [optional]

<!-- Relevant interpretability work for the model goes here -->

[More Information Needed]

Environmental Impact

<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->

Carbon emissions can be estimated using the Machine Learning Impact calculator presented in Lacoste et al. (2019).

• Hardware Type: NVIDIA RTX A6000
• Cloud Provider: [More Information Needed]
• Compute Region: [More Information Needed]
• Carbon Emitted: [More Information Needed]

Technical Specifications [optional]

Model Architecture and Objective

[More Information Needed]

Compute Infrastructure

[More Information Needed]

Hardware

[More Information Needed]

Software

[More Information Needed]

Citation [optional]

<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->

BibTeX:

[More Information Needed]

APA:

[More Information Needed]

Glossary [optional]

<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->

[More Information Needed]

More Information [optional]

[More Information Needed]

Model Card Authors [optional]

[More Information Needed]

Model Card Contact

[More Information Needed]

Framework versions

• PEFT 0.15.0