How to input system message and file prompt to Bloom

🤗Transformers
1

Hi.

This is a function that basically uses the OpenAI API to take an input system message, file prompt and generate an output to eb parsed as a knowledge graph:

def process_gpt(file_prompt, system_msg):
    completion = openai.ChatCompletion.create(
        engine=openai_deployment,
        max_tokens=15000,
        temperature=0,
        messages=[
            {"role": "system", "content": system_msg},
            {"role": "user", "content": file_prompt},
        ],
    )
    nlp_results = completion.choices[0].message.content
    sleep(8)
    return nlp_results

I am trying to do the same here in Bloom:

def process_bloom(system_msg):
    try:
        generator = pipeline('text-generation', model = model, tokenizer=tokenizer)
        nlp_results = generator(system_msg, max_length = 2000)
        nlp_results = nlp_results[0]['generated_text']
        return nlp_results
    except Exception as e:
        print(f'Error processing bloom: (e)')
        return '{"error": Failed to process bloom}'

But I’m using the Hugginface API, not OpenAI, so I couldn’t figure out how to map the exact logic to make use of Bloom.