feat(MODEL-CLM): Contrastive-LM bi-encoder decision model - #3317
Closed
mudler-agent wants to merge 2 commits into
Closed
mudler-agent wants to merge 2 commits into
mudler-agent wants to merge 2 commits into
Conversation
…-DECIDE specs Add five SystemOne-class decision model specs to .agents/specs/: - MODEL-CLM: frozen Qwen3-8B encoder + dual projection heads (state head + action head), scaled-cosine scoring, /v1/systemone - MODEL-TEV1: autoregressive decision model (Qwen3.5-4B SFT), generates option letter via /v1/chat/completions (not /v1/systemone) - MODEL-XOR: 35B MoE (Qwen3.6-A3B) + multimodal (up to 8 images), forward+reverse option-order eval, SGLang oracle - MODEL-JEV: structured generation for DiffusionGemma (vLLM PR #57250), canvas seeding, read-only requests, pinned positions, logprobs - MODEL-GLINER25-DECIDE: DeBERTa-v3-large + classification head (Linear→ReLU→Linear), reuses existing DeBERTa v2 encoder Each spec follows the kev.md template structure. No implementation code — specs only, per AGENTS.md §115 (spec before code).
… dual MLP heads) CLM (Contrastive-LM/CLM-v0.1-8B) is a bi-encoder System 1 decision model: a frozen Qwen3-8B backbone (last-token pooling) feeds two MLP projection heads (state head + action head), each Linear(H,P)→GELU→LayerNorm(P)→ Linear(P,P)→GELU→Linear(P,E), L2-normalized and scored via scaled cosine similarity (exp(logit_scale).clamp(max=100)). Architecture: N+1 forward passes (1 state + N candidates), each a single-sequence prefill through Qwen3DenseModel::ForwardHidden. Confidence formula DIFFERS from kev/laya: max(0, min(1, p_max - mean(rest))) with NO probability rounding. New files: include/vllm/model_executor/models/clm.h — head weights, params, forward decls include/vllm/model_executor/models/clm_inference.h — ClmDecisionResult + ClmInference() src/vllm/model_executor/models/clm_head.cpp — host MLP head forward, L2 norm, scoring src/vllm/model_executor/models/clm_registry.cpp — factory, load, prepare, forward, inference, registration tests/vllm/models/test_clm.cpp — 13 tests, 42 assertions (all pass) Modified: CMakeLists.txt — add clm_head.cpp, clm_registry.cpp tests/CMakeLists.txt — add test_clm src/capi/vllm_c.cpp — allowlist ClmModel + ClmInference dispatch src/vllm/entrypoints/openai/server_main.cpp — ClmModel decision dispatch block src/vllm/entrypoints/openai/systemone.cpp — BuildSystemOneAnswerClm (CLM confidence) include/vllm/entrypoints/openai/systemone.h — declare BuildSystemOneAnswerClm Spec: .agents/specs/clm.md Issue: .agents/issues/MODEL-CLM/ISSUE-LOCAL-01M3AAH2476PHATNMMV36B3FHC.md FOLLOWING_AGENTS_PROTOCOL Following-Agents-Protocol: true AI-Assisted: true Assisted-by: AGENT:glm5.2 [nib]
Collaborator
Author
|
Closing: the model implementation, tests, and docs from this PR were merged to main via #3324 (feat/parakeet-diarization). All model files are on main: clm_head.cpp, clm_registry.cpp, gliner25_decide_head.cpp, gliner25_decide_registry.cpp, xor_registry.cpp, tev1_registry.cpp, and all test files. E2E verified: CLM, GLiNER2.5-Decide, and Tev1-4B all load and serve correctly. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
MODEL-CLM: Contrastive-LM bi-encoder decision model
CLM (Contrastive-LM/CLM-v0.1-8B) is a bi-encoder System 1 decision model: a frozen Qwen3-8B backbone (last-token pooling) feeds two MLP projection heads (state head + action head), each
Linear(H,P)→GELU→LayerNorm(P)→Linear(P,P)→GELU→Linear(P,E), L2-normalized and scored via scaled cosine similarity (exp(logit_scale).clamp(max=100)).Architecture
Qwen3DenseModel::ForwardHiddenmax(0, min(1, p_max - mean(rest)))with NO probability rounding"ClmModel"New files
include/vllm/model_executor/models/clm.h— head weights, params, forward declsinclude/vllm/model_executor/models/clm_inference.h—ClmDecisionResult+ClmInference()src/vllm/model_executor/models/clm_head.cpp— host MLP head forward, L2 norm, scoringsrc/vllm/model_executor/models/clm_registry.cpp— factory, load, prepare, forward, inference, registrationtests/vllm/models/test_clm.cpp— 13 tests, 42 assertions (all pass)Modified
CMakeLists.txt— addclm_head.cpp,clm_registry.cpptests/CMakeLists.txt— addtest_clmsrc/capi/vllm_c.cpp— allowlistClmModel+ClmInferencedispatch invllm_decidesrc/vllm/entrypoints/openai/server_main.cpp—ClmModeldecision dispatch blocksrc/vllm/entrypoints/openai/systemone.cpp—BuildSystemOneAnswerClm(CLM confidence)include/vllm/entrypoints/openai/systemone.h— declareBuildSystemOneAnswerClmSpec
.agents/specs/clm.mdFOLLOWING_AGENTS_PROTOCOL
Following-Agents-Protocol: true
AI-Assisted: true
Assisted-by: AGENT:glm5.2 [nib]