Modelpublic

Berk/reasongraph-extractor-0.6b

sourceHugging Faceapache-2.0updated 1mo agoView on Hugging Face

1likes78downloads

Model Card

reasongraph-extractor-qwen3-0.6b

One small LLM for the three reasongraph extraction tasks -- causal cause/effect/signal spans, contradiction detection, and entity extraction -- from a single LoRA adapter over Qwen/Qwen3-0.6B, selected by a task tag and returning strict JSON. This is the conflict-capable, llama.cpp-servable variant: unlike the qwen3.5 extractor, this base runs in llama.cpp today, so it is the model behind reasongraph's self-hosted FineTunedConflictResolver.

—Method: LoRA (r=16, alpha=32, dropout 0.05, all linear layers, 3 epochs, completion-only loss), merged to fp16.

Tasks & prompt format

Prompt with a bare task tag (no chat template) and greedily decode the JSON completion.

Tag	Input	Output
`[causal]`	`[causal] <sentence>`	`{"causal": true	false, "relations": [{"cause": "...", "effect": "...", "signal": "..."	null}]}` (spans verbatim)
`[conflict]`	`[conflict] existing: <fact A>\nnew: <fact B>`	`{"conflict": true	false}`
`[entities]`	`[entities] <sentence>`	`{"entities": ["...", ...]}`

Evaluation (L1)

Metric	value
CNC subtask-2 dev F1 (official scorer)	0.651
Conflict F1 (40 hand pairs, fp16)	0.851
CPU Q4KM causal latency (4 threads)	~930 ms/sentence, ~3871 sentences/hour, ~1.2 GB RAM

llama.cpp serving (conflict resolver)

llama-server -m qwen3-0.6b-multitask-Q4_K_M.gguf -t 4 -c 2048

Per-pair conflict via /v1/completions (or /completion), temperature 0, cache_prompt, with a JSON grammar so the reply is strict yes/no:

root ::= "{\"conflict\": \"" ("true" | "false") "}"

(exact grammar used in production: root ::= "{" ws "\"conflict\"" ws ":" ws ("true"|"false") ws "}"). Prompt = "[conflict] existing: {existing}\nnew: {new}". Served figures: served Q4KM conflict F1 0.851 at 141 ms/pair (median, 4 threads), p90 162 ms (H3).

Files

—model.safetensors -- merged fp16 model (load with transformers).
—qwen3-0.6b-multitask-Q4_K_M.gguf -- 4-bit GGUF for llama.cpp.
—adapter/ -- standalone LoRA adapter (apply on Qwen/Qwen3-0.6B).

License

Apache-2.0, inherited from the Qwen3 base. Causal News Corpus training text is CC0-1.0.