reasongraph-extractor-1.7b

reasongraph-extractor-qwen3-1.7b

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/Qwen3-1.7B, 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.

Evaluation (L1)

llama.cpp serving (conflict resolver)

llama-server -m qwen3-1.7b-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 Q4_K_M conflict F1 0.976 at 274 ms/pair (median, 4 threads), p90 358 ms -- the grammar-constrained decode beats the fp16 free-form parse (H3).

Files

• model.safetensors -- merged fp16 model (load with transformers).
• qwen3-1.7b-multitask-Q4_K_M.gguf -- 4-bit GGUF for llama.cpp.
• adapter/ -- standalone LoRA adapter (apply on Qwen/Qwen3-1.7B).

License

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