Instructions to use Pushkar27/GriceBench-Repair with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Pushkar27/GriceBench-Repair with Transformers:

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

pipe = pipeline("text-generation", model="Pushkar27/GriceBench-Repair")

# Load model directly
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

tokenizer = AutoTokenizer.from_pretrained("Pushkar27/GriceBench-Repair")
model = AutoModelForSeq2SeqLM.from_pretrained("Pushkar27/GriceBench-Repair", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Pushkar27/GriceBench-Repair with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Pushkar27/GriceBench-Repair"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Pushkar27/GriceBench-Repair",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/Pushkar27/GriceBench-Repair

SGLang

How to use Pushkar27/GriceBench-Repair 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 "Pushkar27/GriceBench-Repair" \
    --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": "Pushkar27/GriceBench-Repair",
		"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 "Pushkar27/GriceBench-Repair" \
        --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": "Pushkar27/GriceBench-Repair",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use Pushkar27/GriceBench-Repair with Docker Model Runner:
```
docker model run hf.co/Pushkar27/GriceBench-Repair
```

Pushkar27 commited on Apr 24

Commit

16dc7b8

1 Parent(s): fa4778c

CRITICAL: Remove all escaped underscores from YAML metadata

Browse files

Files changed (1) hide show

README.md +7 -8

README.md CHANGED Viewed

@@ -4,7 +4,7 @@ language:
 license: apache-2.0
 library_name: transformers
 tags:
-- text2text-generation
 - dialogue
 - gricean-maxims
 - cooperative-communication
@@ -16,13 +16,13 @@ datasets:
 - topical-chat
 metrics:
 - bleu
-pipeline_tag: text2text-generation
 base_model: google-t5/t5-base
 model-index:
 - name: GriceBench-Repair
   results:
   - task:
-      type: text2text-generation
       name: Gricean Maxim Violation Repair
     dataset:
       name: Topical-Chat (GriceBench repair validation split, N=401)
@@ -161,14 +161,14 @@ Per-maxim BLEU scores on the repair validation set (N=401):
 **Three-layer degeneracy prevention:**
 1. Violation-type-aware decoding (nucleus sampling for Manner, beam for others)
-2. Post-generation multi-signal filter (punctuation bursts, trigram repetition, exclamation density)
-3. Graceful fallback — returns original with `is_fallback: True` flag if all attempts fail
 ---
 ## Why Relation Violations Use Retrieval
-Relation violations mean the *entire response* is off-topic — there is nothing to edit. T5 in a seq2seq framing can only edit, not generate entirely new content. We route Relation repairs to a FAISS index over 50,000 Topical-Chat responses (MRR > 0.70, Top-1 accuracy > 60%). See the GitHub repo for the full retrieval system.
 ---
@@ -185,8 +185,7 @@ Relation violations mean the *entire response* is off-topic — there is nothing
 ## Limitations & Biases
-- **Hallucination Risk:** Like all seq2seq models, T5 can occasionally introduce factual errors during repair. Always use the "Quality" detector after repair to verify.
-- **Dependency on Context:** Repair quality is heavily dependent on the provided "Context" being accurate and sufficient.
 - **Mode Collapse:** Avoid using beam search for "Manner" repairs.
 ---

 license: apache-2.0
 library_name: transformers
 tags:
+- text-generation
 - dialogue
 - gricean-maxims
 - cooperative-communication
 - topical-chat
 metrics:
 - bleu
+pipeline_tag: text-generation
 base_model: google-t5/t5-base
 model-index:
 - name: GriceBench-Repair
   results:
   - task:
+      type: text-generation
       name: Gricean Maxim Violation Repair
     dataset:
       name: Topical-Chat (GriceBench repair validation split, N=401)
 **Three-layer degeneracy prevention:**
 1. Violation-type-aware decoding (nucleus sampling for Manner, beam for others)
+2. Post-generation multi-signal filter
+3. Graceful fallback with `is_fallback: True` flag
 ---
 ## Why Relation Violations Use Retrieval
+Relation violations mean the *entire response* is off-topic — there is nothing to edit. We route Relation repairs to a FAISS index over 50,000 Topical-Chat responses (MRR > 0.70, Top-1 accuracy > 60%).
 ---
 ## Limitations & Biases
+- **Hallucination Risk:** T5 can occasionally introduce factual errors during repair. Always verify with the "Quality" detector.
 - **Mode Collapse:** Avoid using beam search for "Manner" repairs.
 ---