Setting the file. One moment.
Skill 24 · Train Sentence Transformers
Subchapter 24.10
references/losses_multi_vector_encoder.mdMarkdown10 KBView on GitHub
All losses live in sentence_transformers.multi_vector_encoder.losses.
Multi-vector models emit one embedding per token and score via MaxSim (for each query token, take the max similarity to any document token, then sum across query tokens). This is fundamentally different from single-vector cosine, which changes what “temperature” means: MaxSim is an unbounded sum over query-token similarities, so for it a scale near 1.0 makes sense. With MeanMaxSim scoring (, and set so evaluation matches), each score is divided by its query’s token count, so a scale of roughly the average length is a reasonable start, although some works use larger scales.
(query, document)scale=1.0 (temperature=1.0) is correct for unnormalized MaxSim: do not copy scale=20.0 from bi-encoder MNRL.similarity_fct=mean_colbert_scoresmodel.similarity_fn_name = "meanmaxsim"| You have | Use |
|---|---|
(anchor, positive) or (anchor, positive, negative) triplets | MultiVectorMultipleNegativesRankingLoss |
| Same, want effective batch size of 128+ | CachedMultiVectorMultipleNegativesRankingLoss |
Cross-encoder teacher scores, (query, positive, negative, score_diff) | MultiVectorMarginMSELoss |
Listwise distillation (query, [doc_1..doc_N], scores) | MultiVectorDistillKLDivLoss |
Hard-negative mining is essential for competitive results, because random in-batch negatives leave a lot on the table for late-interaction models. See dataset_formats.md (Hard-negative mining section) and ../scripts/mine_hard_negatives.py.
The default late-interaction contrastive loss. In-batch positives plus explicit hard negatives, scored with MaxSim (or XTR).
from sentence_transformers.multi_vector_encoder.losses import MultiVectorMultipleNegativesRankingLoss
loss = MultiVectorMultipleNegativesRankingLoss(model=model) # scale=1.0 (temperature=1.0) is the correct default(anchor, positive) or (anchor, positive, negative_1, ..., negative_n). The collator stamps each column’s task (column 0 becomes query, others become document). Pass task=... on a column to override.batch_sampler=BatchSamplers.NO_DUPLICATES on training args (same reason as bi-encoder MNRL).similarity_fct=colbert_scores by default. Pass XTRScores(top_k=...) for the sparser XTR-style scoring (from sentence_transformers.multi_vector_encoder.scoring). XTR applies to training only: evaluation always scores with MaxSim (see Gotchas).scale=1.0 (temperature=1.0) matches PyLate and is correct for unnormalized MaxSim. Do NOT copy scale=20.0 from bi-encoder MNRL. With MeanMaxSim (similarity_fct=mean_colbert_scores), raise the scale to roughly the average query length instead, which is the same objective at the same strength.GradCache variant: chunked embedding forward, cached gradients, and a second re-embedding pass. Decouples per-device batch size from effective in-batch negatives.
loss = CachedMultiVectorMultipleNegativesRankingLoss(
model=model,
mini_batch_size=8, # or use mini_batch_num_tokens=... for token-budgeted packing
score_mini_batch_size=4, # optional, smaller trims the transient (Q, Q*N, q_tok, d_tok) buffer
)gradient_checkpointing=True (same as every Cached* loss).mini_batch_num_tokens=N packs sequences until N real tokens per mini-batch instead of a fixed mini_batch_size. Big win on variable-length data with flash-attention / input flattening.score_mini_batch_size chunks the SCORING phase (which builds (Q, Q*N, q_tokens, d_tokens) intermediates) independently from the embedding phase. Drop it first when hitting OOM in the loss stage.chunk_elements is the second lever on that same phase, and it cuts the other axis: score_mini_batch_size chunks queries, chunk_elements chunks documents inside each MaxSim call. Pass it through the scorer, e.g. similarity_fct=partial(colbert_scores, chunk_elements=1_000_000). Scores and gradients are unchanged (it is a pure memory knob), and it composes with score_mini_batch_size.gather_across_devices=True gathers document embeddings across DDP ranks. Enables cross-rank in-batch negatives.Distillation is where multi-vector models learn most efficiently: cross-encoder teachers (e.g. gte-modernbert-base) score (query, doc) pairs offline, and the student MaxSim model regresses to that signal.
For the standard MS MARCO KD format (LightOn’s ms-marco-en-bge and similar: (query_id, document_ids, scores) with separate queries / documents splits), you can use sentence_transformers.util.resolve_ids to join the IDs against the text splits at load time. It’s a factory that returns a batched transform: pass a mapping from each input ID column to its lookup dataset, e.g. dataset.set_transform(resolve_ids({"query_id": queries, "document_ids": documents}, max_list_length=32)) (lazy, no caching) or via dataset.map(..., batched=True, remove_columns=["query_id", "document_ids"]) (eager, cached: map keeps input columns unless told to drop them, so pass the ID columns to remove_columns), yielding (query, document_1, ..., document_32, scores) rows ready for MultiVectorDistillKLDivLoss. The same mapping shape covers triplets with IDs ({"query_id": queries, "positive_id": documents, "negative_id": documents}) and ID-only contrastive datasets without a scores column.
Regress the margin between positive and negative MaxSim scores against the teacher’s margin.
loss = MultiVectorMarginMSELoss(model=model)(query, positive, negative, score_diff) where score_diff = teacher_score(query, positive) - teacher_score(query, negative). Raw teacher scores [score(q, pos), score(q, neg_1), ...] also work, converted to margins internally.similarity_fct convention as the other losses, but pairwise-shaped: defaults to colbert_scores_pairwise, or pass xtr_scores_pairwise for XTR.Listwise KL-div: student’s softmax distribution over N candidates should match the teacher’s.
loss = MultiVectorDistillKLDivLoss(model=model)(query, document_1, ..., document_N, scores) where scores is a list of N teacher scores per row. One column per candidate document (the standard positional multi-column convention), and the label column must use a recognized name (label, labels, score, scores). resolve_ids expands a stored ID list into the numbered columns.MarginMSE when you have full N-way teacher scores (not just positive/negative margins).similarity_fct defaults to colbert_kd_scores, the listwise KD variant, not the colbert_scores used by the MNRL family. XTRKDScores is the XTR counterpart.temperature softens both distributions before the softmax, and can be overridden per side with student_temperature / teacher_temperature. Set teacher_temperature from the spread of your teacher’s scores rather than from a value copied out of a paper: a float32 softmax underflows to exact zeros once a row’s spread divided by the temperature exceeds about 100, and every candidate that underflows drops out of the KL, which is the ranking information you are distilling. Sharpening student_temperature far below the student’s own score spread collapses that side too, and once both are one-hot the loss and its gradient are exactly zero. Start at 1.0 / 1.0 and only move a side you have a measured reason to move.student_temperature=0.001: when combining KD with a contrastive loss, scale the KD weight back up accordingly.per_device_train_batch_size first and raise gradient_accumulation_steps to hold the effective batch. Only reduce N (candidate-list length) as a last resort, since lowering N changes the experiment.self.model(sf, task="query" if idx==0 else "document"). Use the collator’s router_mapping to override if you need a non-standard column layout.T is handled: the batch’s positive and negative columns can have different token counts (Qwen2-VL family etc.). stack_padded_token_embeddings pads to the per-batch max before MaxSim.MultiVectorMask rewrites the input mask). Custom loss code that reads sentence_features[i]["attention_mask"] directly instead of outputs[i]["attention_mask"] will silently score against skiplisted tokens.scale=20.0 copied from bi-encoder MNRL: misscaled for the unnormalized MaxSim sum. Keep scale=1.0 there. Only length-normalized MeanMaxSim wants a larger scale, roughly the average query length.Normalize at the token level in the pipeline: colbert_scores assumes L2-normalized token embeddings. If your custom pipeline drops the token-level Normalize, either add one or pass normalize_embeddings=True semantics via a wrapper.CachedMultiVectorMultipleNegativesRankingLoss + gradient_checkpointing=True: crash. Pick one.MultiVectorMarginMSELoss without precomputed score_diff: label column must be populated from a teacher pass ahead of training. The loss does not compute the teacher inline.XTRScores is a train-only similarity_fct. XTR takes its top-k across the whole candidate set, so a (query, document) pair has no standalone score, which means it cannot be the model’s similarity_fn_name and the evaluators reject it outright. Evaluation and inference score with MaxSim, also for XTR-trained models. That is by design, not a mismatch to fix. To score a fixed candidate set ad hoc, call xtr_scores directly.gte-modernbert-base, mxbai-rerank-large-v2) for the teacher pass.