Instructions to use jackal79/gpt2-ibce-lowrank-192 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use jackal79/gpt2-ibce-lowrank-192 with Transformers:

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

pipe = pipeline("text-generation", model="jackal79/gpt2-ibce-lowrank-192", trust_remote_code=True)

# Load model directly
from transformers import AutoTokenizer, IBCEGPT2LowRank

tokenizer = AutoTokenizer.from_pretrained("jackal79/gpt2-ibce-lowrank-192", trust_remote_code=True)
model = IBCEGPT2LowRank.from_pretrained("jackal79/gpt2-ibce-lowrank-192", trust_remote_code=True)

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use jackal79/gpt2-ibce-lowrank-192 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "jackal79/gpt2-ibce-lowrank-192"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "jackal79/gpt2-ibce-lowrank-192",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/jackal79/gpt2-ibce-lowrank-192

SGLang

How to use jackal79/gpt2-ibce-lowrank-192 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 "jackal79/gpt2-ibce-lowrank-192" \
    --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": "jackal79/gpt2-ibce-lowrank-192",
		"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 "jackal79/gpt2-ibce-lowrank-192" \
        --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": "jackal79/gpt2-ibce-lowrank-192",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use jackal79/gpt2-ibce-lowrank-192 with Docker Model Runner:
```
docker model run hf.co/jackal79/gpt2-ibce-lowrank-192
```

GPT-2 + IBCE Low-Rank (trust_remote_code)

What it is. A runtime wrapper that keeps GPT-2’s fused attention kernels and swaps only QKV/PROJ linears to low-rank factors.
No weights are stored here: on load it fetches openai-community/gpt2 and patches it at runtime. Because this uses trust_remote_code, HF Inference API/Serverless is not supported; run in your own environment or notebook.

Headline results (A100, FP16, greedy): ~1.8–2.0x tokens/sec at ~0.8% perplexity delta on WikiText-2 (raw).

Quick start (pin a safe revision)

from transformers import AutoTokenizer, AutoModelForCausalLM
import torch as th

repo = "jackal79/gpt2-ibce-lowrank-192"
rev  = "a71015d1333757bedaaa2437333cd44415076218"  # pinned revision

tok = AutoTokenizer.from_pretrained(repo, trust_remote_code=True, revision=rev)
mdl = AutoModelForCausalLM.from_pretrained(
    repo, trust_remote_code=True, revision=rev, dtype=th.float16
).cuda().eval()

x = tok("Hello from IBCE low-rank!", return_tensors="pt").to("cuda")
y = mdl.generate(**x, max_new_tokens=64, do_sample=False)
print(tok.decode(y[0], skip_special_tokens=True))

## Licensing
This repository is released under an **Evaluation License** for non-production, non-hosted use.
Commercial production or hosted use requires a separate license.
Contact **jack@thestardrive.com**.

The full license text is in [`LICENSE.txt`](https://huggingface.co/jackal79/gpt2-ibce-lowrank-192/blob/main/LICENSE.txt).

Downloads last month: 2

Model tree for jackal79/gpt2-ibce-lowrank-192

Base model

openai-community/gpt2

Finetuned

(2181)

this model

Evaluation results

Perplexity (GPT-2 baseline) on WikiText-2 (raw)
self-reported

51.641
Perplexity (IBCE) on WikiText-2 (raw)
self-reported

52.068
Tokens per second (baseline, A100 FP16, 512 tok) on WikiText-2 (raw)
self-reported

102.000
Tokens per second (IBCE, A100 FP16, 512 tok) on WikiText-2 (raw)
self-reported

190.000