Fix some glitches in extract_features.py #21

cnrpman · 2018-11-15T07:49:20Z

Do the following fixing to make the extract_features.py runnable:

Add no_cuda argument
Fix the "not all arguments converted during string formatting" error thrown at line 230

1. Add no_cuda argument 2. Fix the "not all arguments converted during string formatting" error thrown at line 230

thomwolf · 2018-11-17T22:07:20Z

Thanks, I've pushed these fixes in the first release (the organization of the library changed quite a bit).

Add ORT internal options config flag

j

[whisper] Add support for audio clips > 30 seconds

expand the first dimension of input_semantics to batch size

* move `TestAssistedCandidateGeneratorDifferentTokenizers` into a new testing file * refactor * NOTHING. add space to rerun github actions tests * remove it... * `UniversalSpeculativeDecodingGenerator` * Use `UniversalSpeculativeDecodingGenerator` when `generation_config.do_sample=True` * assistant tokenizes only the target's new suffix * formatting * fix code * fix code * formatting * add `TestGenerateWithDifferentModels` * `TestGenerateWithDifferentModels` parameterize on `do_sample` * `AssistantVocabMapping` & `AssistantVocabMappingCache` * formatting * `AssistantToTargetTranslator`: `get_target_input_ids` & `get_target_logits` * improve `_get_assistant_to_target_input_ids` & formatting * renaming * WIP: debugging `min_new_tokens` * fix get_target_ids * `UniversalSpeculativeDecodingGenerator` * assistant tokenizes only the target's new suffix * formatting * fix code * fix code * formatting * `TestGenerateWithDifferentModels` parameterize on `do_sample` * `AssistantVocabMapping` & `AssistantVocabMappingCache` * formatting * `AssistantToTargetTranslator`: `get_target_input_ids` & `get_target_logits` * improve `_get_assistant_to_target_input_ids` & formatting * renaming * WIP: debugging `min_new_tokens` * fix get_target_ids * fix device issue * fix get_assistant_input_ids * add `TestAssistedCandidateGeneratorDifferentTokenizers` * formatting * `AssistantVocabTranslatorCache` refactor & tests * revert changes in `src/transformers/generation/logits_process.py` * refactor `AssistedCandidateGenerator` * refactor `AssistedCandidateGeneratorDifferentTokenizers` * formatting * refactor `UniversalSpeculativeDecodingGenerator` * fix negative value for max_new_tokens * fix generation length target + attention_mask vs. assistant + attent * fix device * fix negative max_new_tokens bug * fix UAG * minor * formatting * `AssistedCandidateGeneratorDifferentTokenizers` `lookbehind`s init * resolve conflict & formatting * rerun CI tests * remove space... * remove old code * fix candidate_input_ids device * minor * formatting * Fix prepare + apply (#7) * fix prepare + apply * move to cpu * simplity suppress_tokens * fix bugs and refacatoring * device move * handle self.config.vocab_size > len(target_tokenizer.get_vocab()) * no need to normalize in candidate_generator * address Nadav's comments + minor * optimize device move + SuppressTokensLogitsProcessor * AssistantToTargetTranslator, SuppressTokensLogitsProcessor and tokenizers mapping improvements * padding size * padding improvement * fix and simplify get_target_logits * renaming in get_target_logits * minor * add filter_value and suppress_tokens_id * style + rename * remove TODO * restore original SelectTokensLogitsProcessor with modification * fix style * fix _update_past_and_masks and optimize code * remove assistant_vocab_size arg * fix attention_mask * call _prepare_attention_mask also if not has_past_key_values * handling attention mask for first generation * comment * restore test * remove SelectTokensLogitsProcessor * _update_past_and_masks implementation for USD * Add unittests for Universal Assisted generation * fix style * update tests * Remove unused import and fix `test_speculation_depth` test * exclude special and reserved tokens from tokenizer for UAG * mv `test_universal_assisted_generation.py` to `generation/test_candidate_generator.py` * Remove unused imports and fix style using `make style` (#9) * formatting * Swap gated `meta-llama/llama-3.2` with `allenai/llama` (#10) * Fix space sign disagreement (#12) * default values for AssistantToTargetTranslator fileds * fix space sign * minor * fix test + style * Default values for some fields of assistant to target translator (#11) * default values for AssistantToTargetTranslator fileds * fix * add support to empty logit_processors * Update candidate_generator.py (#15) fix typo * BUG fix in _prepare_assistant_input_ids (#14) * fix _prepare_assistant_input_ids * target_to_assistant_input_ids * Update src/transformers/generation/candidate_generator.py Co-authored-by: Nadav Timor <nadav.timor@weizmann.ac.il> --------- Co-authored-by: Nadav Timor <nadav.timor@weizmann.ac.il> * typo (`target_to_assistant_input_ids`) * formatting * merge upstream/main * Fix minor review comments (#16) * Fix: `token_ids.to(torch.int64)` (#18) * tok ids to `torch.int64` (reference: https://huggingface.co/docs/transformers.js/en/api/tokenizers) * `LongTensor` * fix dtype * `assistant_input_ids.to(dtype=torch.long)` * Remove unused import from test_candidate_generator.py * Remove unused import from test_candidate_generator.py * Remove `numpy` import * resolve pr comments (#19) * `AssistantToTargetTranslator` docstring * (per gante's comment) `filter_value` and `suppress_tokens_id` to class constants * update `AssistantToTargetTranslator` docstring * (gante's comment) replace `match-case` * formatting * Fix Joao's comments (#21) * remove threading * fix logits_processor * fix test device * fix style (#23) * Move atm (#24) * move AssistantToTargetTranslator * fixup * fix logit_processor * add atm_translator test * refactor test * remove threading from test * add require_torch in tests * move AssistantVocabTranslatorCache + add tests * ruff fix --------- Co-authored-by: jmamou <jonathan.mamou@intel.com> Co-authored-by: Gaurav <gauravj@d-matrix.ai> Co-authored-by: Gaurav Jain <gaurjain14@gmail.com> Co-authored-by: gauravjain14 <41287729+gauravjain14@users.noreply.github.com>

* remove threading * fix logits_processor * fix test device

Adding fbgemm

* initial commit of Gemma 3n scaffold * Fixing param pass through on Gemm3p5RMSNorm * Adds Einsum layer to Gemma 3n * Updating EinsumLayer API * Undoing erroneous force push * Reverting RMSNorm to with_scale by default * Adds LAuReL to Gemma 3n * Adds AltUp to Gemma 3n * Adding Gemma3p5 overall and text config with vision and audio config placeholders (huggingface#3) * Adding gemma3p5 text configs * Adding audio config placeholders * Adding a placeholder for vision configs * Updating MobileNetVisionConfig, inheriting TimmWrapperConfig * Updating text configs * Update src/transformers/models/gemma3p5/modular_gemma3p5.py Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Removing altup configs to accept the suggested configs * Update src/transformers/models/gemma3p5/modular_gemma3p5.py Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Updating altup config * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Addressing review comments and updating text configs * Adding a config for activation sparsity * Updating configs to pass through options to super class init and adjust some name prefixes * Updating laurel and altup with corrected config values * Normalizing sub_config initializers --------- Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Updating MLP with activation sparsity (#2) * Updating DecoderBlock for Gemma 3n (huggingface#3) * Initial Gemm3nTextModel (huggingface#4) NOTE: This implementation WILL CHANGE in the coming weeks, however, changes will be strictly additive and this will remain a suitable baseline for downstream implementations to reference. * Adding KV Cache Sharing * Adds Einsum layer to Gemma 3n * Updating EinsumLayer API * Refactored kv cache sharing in attention * Adding KVStore for cache sharing * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update src/transformers/cache_utils.py Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Undoing erroneous force push * Reverting RMSNorm to with_scale by default * Adds LAuReL to Gemma 3n * Updating KV Cache Sharing implementation * Updating the q and k norm definitions in the attention module * Fixing name error for q,k,v RMS norm to use the right 3n module * Updating MLP with activation sparsity * Updating DecoderBlock for Gemma 3.5 * Updating kv cache sharing implementation with the use of a cache buffer and refactoring some lines of code * Isolating KV Cache logic to relevant components * Fixing logic error in Gemma3nAttention.forward * Refactoring caching contributions and fixing kv_store initialization * Simplifying Configs * Remove errant self from super init call * Bug fix in the Attention module - changing self.head_dim to config.head_dim * Bug fixes in the LaurelBlock and RMS Norm super init call * removing redundant code from a merge * Adding per_layer_inputs to TextModel * Adding preprocess embeddings with altup * Adds per-layer-to-single output and a host of TODOs * Integrating altup predict with the model workflow and other minor bug fixes * Using nn.Embedding temporarily for text model * It goes forward * Minor refactor of attention sparsity and RoPE initialization * Fixing duplicate rope_scaling param bug when loading from pretrained --------- Co-authored-by: Sindhu Raghuram <sindhuraghuram@google.com> Co-authored-by: SindhuRaghuram97 <114270661+SindhuRaghuram97@users.noreply.github.com> * Normalizing on altup_num_inputs config option * regenerating modeling file after syncing to HEAD * Use torch.std(..., unbiased=False) for activation sparsity (huggingface#8) * Refactoring to a single QVK Norm (huggingface#13) * AltUp: support scale_corrected_output (huggingface#14) * Converts einsums to nn.Linear (huggingface#7) * Converts einsums to nn.Linear * Removing unused variables * Aligning SharedKVCache with HybridCache (huggingface#11) * Alinging SharedKVStore with HybridCache * Remove KVStore. Refactor apply_rotary_pos_emb for sharing * Addressing review comments * Supporting split modality embeddings in Gemma3n (huggingface#10) * Adding the Embedder class * Update modular Co-authored-by: Ryan Mullins <ryan@ryanmullins.org> * Update modular Co-authored-by: Ryan Mullins <ryan@ryanmullins.org> * Update modular Co-authored-by: Ryan Mullins <ryan@ryanmullins.org> * Update modular Co-authored-by: Ryan Mullins <ryan@ryanmullins.org> * Update modular Co-authored-by: Ryan Mullins <ryan@ryanmullins.org> * Update modular Co-authored-by: Ryan Mullins <ryan@ryanmullins.org> * Addressing review comments, adding audio embedding layers, integrating embedder with the remaining architecture, adding a forward method for conditional generation * Apply suggestions from code review Co-authored-by: Ryan Mullins <ryan@ryanmullins.org> * Update modular Co-authored-by: Ryan Mullins <ryan@ryanmullins.org> * Addressing review comments, prop drilling audio and vision configs to the text config * Removing TODO's that have been addressed * Simplify Embedder init and add audio embeddings * Embeddings refactor. Adds Gemma3nAudioEmbedder and Gemma3nVisionEmbedder * Refactoring vision and audio embeddings into ConditionalGeneration model --------- Co-authored-by: Ryan Mullins <ryan@ryanmullins.org> Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Updating attention mask for Gemma 3.5 (huggingface#15) * xxx_token_index to xxx_token_id * remvoing deprecated last_cache_position * Removing references to SigLIP * Always init per-layer inputs * Using torch.finfo().min for epsilon_tensor * Gemma3nDecoderLayer inherits from Gemma3DecoderLayer. Remove gating lambdas * fix modular GEMMA3N_INPUTS_DOCSTRING * Gemma3nAttention inherits from Gemma3Attention * Modular inheritance fixes * CausalLM conversion script for 4B model (huggingface#16) * Add Gemma3n Audio Encoder (huggingface#6) * initial commit of Gemma 3.5 scaffold * Fixing param pass through on Gemm3nRMSNorm * Adds Einsum layer to Gemma 3.5 * Updating EinsumLayer API * Undoing erroneous force push * Reverting RMSNorm to with_scale by default * Adds LAuReL to Gemma 3n * Adds AltUp to Gemma 3n * Adding Gemma3n overall and text config with vision and audio config placeholders (huggingface#3) * Adding gemma3n text configs * Adding audio config placeholders * Adding a placeholder for vision configs * Updating MobileNetVisionConfig, inheriting TimmWrapperConfig * Updating text configs * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Removing altup configs to accept the suggested configs * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Updating altup config * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Addressing review comments and updating text configs * Adding a config for activation sparsity * Updating configs to pass through options to super class init and adjust some name prefixes * Updating laurel and altup with corrected config values * Normalizing sub_config initializers --------- Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Updating MLP with activation sparsity (#2) * Updating DecoderBlock for Gemma 3.5 (huggingface#3) * Initial Gemm3nTextModel (huggingface#4) NOTE: This implementation WILL CHANGE in the coming weeks, however, changes will be strictly additive and this will remain a suitable baseline for downstream implementations to reference. * Adding KV Cache Sharing * Adds Einsum layer to Gemma 3.5 * Updating EinsumLayer API * Refactored kv cache sharing in attention * Adding KVStore for cache sharing * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update src/transformers/cache_utils.py Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Undoing erroneous force push * Reverting RMSNorm to with_scale by default * Adds LAuReL to Gemma 3n * Updating KV Cache Sharing implementation * Updating the q and k norm definitions in the attention module * Fixing name error for q,k,v RMS norm to use the right Gemma 3n module * Updating MLP with activation sparsity * Updating DecoderBlock for Gemma 3.5 * Updating kv cache sharing implementation with the use of a cache buffer and refactoring some lines of code * Isolating KV Cache logic to relevant components * Fixing logic error in Gemma3nAttention.forward * Refactoring caching contributions and fixing kv_store initialization * Simplifying Configs * Remove errant self from super init call * Bug fix in the Attention module - changing self.head_dim to config.head_dim * Bug fixes in the LaurelBlock and RMS Norm super init call * removing redundant code from a merge * Adding per_layer_inputs to TextModel * Adding preprocess embeddings with altup * Adds per-layer-to-single output and a host of TODOs * Integrating altup predict with the model workflow and other minor bug fixes * Using nn.Embedding temporarily for text model * It goes forward * Minor refactor of attention sparsity and RoPE initialization * Fixing duplicate rope_scaling param bug when loading from pretrained --------- Co-authored-by: Sindhu Raghuram <sindhuraghuram@google.com> Co-authored-by: SindhuRaghuram97 <114270661+SindhuRaghuram97@users.noreply.github.com> * Normalizing on altup_num_inputs config option * Adding audio encoder config * Adds high-level components for Audio Encoder * Implement uniform reducer for Audio Encoder * Adding placeholders for Conformer components in Audio Encoder * Adding placeholders for SubSampleConvProjection components in Audio Encoder * Adding SequenceLayer component placeholders * Implementing Gemma3nAudioEncoder with nn.Sequential * Implementing Gemma3nAudioSubSampleConvProjection with nn.Sequential * Implementing Conformer model with SequenceLayers * Use OrderedDict in nn.Sequential initializers * Implements sl.Residual in Torch with nn.Sequential and OrderedDict * Adopting a base SequenceLayer class with default forward() method * Implementing sl.GatedLinearUnit in Torch * Implementing sl.Swish in Torch * Implementing sl.ReLU in Torch * Implementing sl.Scale in Torch * Removing sl.Dropout after tree-shaking * Implementing sl.RMSNorm in Torch with fake shape * Implementing sl.GroupNorm in Torch * Implementing sl.Conv2d in Torch * Implementing sl.Dense in Torch * Removing sl.Delay layers, which act as pass-throughs * Connecting shapes to configs in initializers * Removing sl.Emit * Implementing sl.ExpandDims in Torch * Adding sl.GradientClipping to Torch * Implementing sl.DenseShaped in Torch * Implementing sl.LDPA in Torch * Removing unused sl.CombinedQKVProj class * Fixing erroneous type hint * Implemnenting sl.DepthwiseConv1D in Torch * Implementing sl.MaskInvalid in Torch * Fixes for initialization * Fixes for saving weights * Removing einsums per feedback from HF staff * Removing Sequence Layers idioms from audio encoder * Fixes for reviewer comments * CausalLM conversion script for 4B model * inv_timescales to non-persistent buffer * Addressing audio encoder Attention feedback * Addressing Gemma3nAudioSSCPConvBlock feedback * Addressing Gemma3nAudioConformerAttention feedback * Addressing padding feedback * Weights conversion loads audio state dict * Always use vision_config so saving works * Token id updates for configs * Stubs for interleaving audio embs * Addressing reviewer feedback --------- Co-authored-by: SindhuRaghuram97 <114270661+SindhuRaghuram97@users.noreply.github.com> Co-authored-by: Sindhu Raghuram <sindhuraghuram@google.com> * Fixing cache access error * Removing duplicate code from a bad merge * Gemma 3n Text + Vision Part 1 (huggingface#17) * testing utilities for numerics comparisons * Corrected einsum to nn.Linear weights conversion * Inherit scaled word embs from Gemma3 not Bart * Fixing transposes for collapsed linears * More transpose fixes * numpy api fix * RMSNorm: Explicit kwargs, scale_shift=0.0 when with_scale=True * Force AltUp to float32 * Updating debugging script for AudioEncoder debugging * Support divide_weight_by_sqrt_fan_in from JAX for per-layer inputs * Correcting attention einsum conversions * RMSNorm in type of x * Fixing douplicate laurel norm/gating * KV sharing using the right previous indices * Refactor kv shared index computation. Correct frac_shared_layers * Use num_shared_layers instead of inferring from a fraction * fixing a bug for logging * Fix shared data_ptrs in altup inits * rope: adjust proj -> norm -> rope to preserve computation (huggingface#20) * rope: adjust proj -> norm -> rope to preserve computation * Removing some breaking language model fluff in ConditionalGeneration * Consolidate query_states transforms --------- Co-authored-by: Douglas Reid <21148125+douglas-reid@users.noreply.github.com> Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Vectorize the loops in AltUp (huggingface#19) * Vectorize the loops in AltUp * fix typo * Expanding to support batched inputs * remove extra debug script * Fix AltUp.forward --------- Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Add 'scale_shift=0.0, with_scale=True' to the final norm in TextModel * Convert norm to 1/sqrt (huggingface#21) * Convert norm to 1/sqrt * Scale shift change per Phil's rec * Adding default activation sparsity * Fixing 2B config in weights conversion script * Fixing RMSNorm parameters - adding scale_shift and with_scale * Correcting query pre-attention scaling * Adding query_rescale_scalar to text config * Adding layer_idx to MLP * Permafix for input_layernorm * Use 1/sqrt instead of rsqrt in DecoderLayer * Fix o_proj conversion * Conversion script update for vision encoder * Removing logging for debugging timm model * Fixing bugs in Gemma3nForConditionalGeneration for text generation * Generating the modeling_gemma3n.py file * Removing the addition of an erroneous line in the modeling file * Adding gemma3n text model to modeling_auto * Bugfix: Updating the interleaving of inputs_embeds and vision_embeds * Updating the modeling file with the latest bugfix changes * Updating models/auto for Gemma 3n * using AutoTokenizer in forward test * Adding processing_gemma3n.py * Gemma 3n configured for AutoModel. Conversion script updated. * Removing errant merge artifacts --------- Co-authored-by: Mayank Chaturvedi <imayank@google.com> Co-authored-by: Douglas Reid <douglas-reid@users.noreply.github.com> Co-authored-by: Douglas Reid <21148125+douglas-reid@users.noreply.github.com> Co-authored-by: Xuan-Son Nguyen <thichthat@gmail.com> Co-authored-by: Sindhu Raghuram <sindhuraghuram@google.com> * Removing errant debugging statements from Gemma 3 * Gemma3n audio model (huggingface#18) * testing utilities for numerics comparisons * Implement CumulativeGroupNorm and add to SubSampleConvProjection and SSCPConvBlock * Add audio version of forward script based on RyanMullins' implementation * Updating to match encoder tests. WIP: config question needs resolving * Updates to audio classes to enable end-to-end running * Removing vestigial classes, cleaning up print statements * Adding SiLU / Swish to audio conformer feed forward block * Shifted Gemma3p5Audio naming prefix to Gemma3NanoAudio * Adding outputs to audio test * Fixes to padding in SSCP and 1D convolution, align RMS Norm with wider model * Update forward test to load from local weights * Update conversion to process / output audio layers * Update __all__ to export audio encoder * AutoModel registration for Gemma 3n Audio * Use AutoModel for ConditionalGeneration.audio_tower * Fixing input_proj_linear transpose * Fixing Gemma3NanoAudioConformerAttention.post conversion * Fixing Gemma3NanoAudioSSCPConvBlock.conv weights conversion * Correcting indentation issue on Gemma3p5RMSNorm --------- Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Text + Vision Part 2 (huggingface#23) * Updates for ConditionalGeneration.get_image_features * Adding a WIP draft of image_processing_gemma3p5.py * Update src/transformers/models/gemma3p5/modular_gemma3p5.py Co-authored-by: SindhuRaghuram97 <114270661+SindhuRaghuram97@users.noreply.github.com> * Modular conversion after github suggested change * Text + image gives good results * Fixing image size preset * Updating configs for the 2B variant in the conversion script * Using final generation config in conversion script --------- Co-authored-by: Sindhu Raghuram <sindhuraghuram@google.com> Co-authored-by: SindhuRaghuram97 <114270661+SindhuRaghuram97@users.noreply.github.com> * Audio Integration (huggingface#12) * initial commit of Gemma 3n scaffold * Fixing param pass through on Gemm3nRMSNorm * Adds Einsum layer to Gemma 3n * Updating EinsumLayer API * Undoing erroneous force push * Reverting RMSNorm to with_scale by default * Adds LAuReL to Gemma 3n * Adds AltUp to Gemma 3n * Adding Gemma 3n overall and text config with vision and audio config placeholders (huggingface#3) * Adding Gemma 3n text configs * Adding audio config placeholders * Adding a placeholder for vision configs * Updating MobileNetVisionConfig, inheriting TimmWrapperConfig * Updating text configs * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Removing altup configs to accept the suggested configs * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Updating altup config * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Addressing review comments and updating text configs * Adding a config for activation sparsity * Updating configs to pass through options to super class init and adjust some name prefixes * Updating laurel and altup with corrected config values * Normalizing sub_config initializers --------- Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Updating MLP with activation sparsity (#2) * Updating DecoderBlock for Gemma 3n (huggingface#3) * Initial Gemma3nTextModel (huggingface#4) NOTE: This implementation WILL CHANGE in the coming weeks, however, changes will be strictly additive and this will remain a suitable baseline for downstream implementations to reference. * Adding KV Cache Sharing * Adds Einsum layer to Gemma 3n * Updating EinsumLayer API * Refactored kv cache sharing in attention * Adding KVStore for cache sharing * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update src/transformers/cache_utils.py Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Undoing erroneous force push * Reverting RMSNorm to with_scale by default * Adds LAuReL to Gemma 3n * Updating KV Cache Sharing implementation * Updating the q and k norm definitions in the attention module * Fixing name error for q,k,v RMS norm to use the right 3n module * Updating MLP with activation sparsity * Updating DecoderBlock for Gemma 3n * Updating kv cache sharing implementation with the use of a cache buffer and refactoring some lines of code * Isolating KV Cache logic to relevant components * Fixing logic error in Gemma3nAttention.forward * Refactoring caching contributions and fixing kv_store initialization * Simplifying Configs * Remove errant self from super init call * Bug fix in the Attention module - changing self.head_dim to config.head_dim * Bug fixes in the LaurelBlock and RMS Norm super init call * removing redundant code from a merge * Adding per_layer_inputs to TextModel * Adding preprocess embeddings with altup * Adds per-layer-to-single output and a host of TODOs * Integrating altup predict with the model workflow and other minor bug fixes * Using nn.Embedding temporarily for text model * It goes forward * Minor refactor of attention sparsity and RoPE initialization * Fixing duplicate rope_scaling param bug when loading from pretrained --------- Co-authored-by: Sindhu Raghuram <sindhuraghuram@google.com> Co-authored-by: SindhuRaghuram97 <114270661+SindhuRaghuram97@users.noreply.github.com> * Normalizing on altup_num_inputs config option * Adding audio encoder config * Adds high-level components for Audio Encoder * Implement uniform reducer for Audio Encoder * Adding placeholders for Conformer components in Audio Encoder * Adding placeholders for SubSampleConvProjection components in Audio Encoder * Adding SequenceLayer component placeholders * Implementing Gemma3nAudioEncoder with nn.Sequential * Implementing Gemma3nAudioSubSampleConvProjection with nn.Sequential * Implementing Conformer model with SequenceLayers * Use OrderedDict in nn.Sequential initializers * Implements sl.Residual in Torch with nn.Sequential and OrderedDict * Adopting a base SequenceLayer class with default forward() method * Implementing sl.GatedLinearUnit in Torch * Implementing sl.Swish in Torch * Implementing sl.ReLU in Torch * Implementing sl.Scale in Torch * Removing sl.Dropout after tree-shaking * Implementing sl.RMSNorm in Torch with fake shape * Implementing sl.GroupNorm in Torch * Implementing sl.Conv2d in Torch * Implementing sl.Dense in Torch * Removing sl.Delay layers, which act as pass-throughs * Connecting shapes to configs in initializers * Removing sl.Emit * Implementing sl.ExpandDims in Torch * Adding sl.GradientClipping to Torch * Implementing sl.DenseShaped in Torch * Implementing sl.LDPA in Torch * Removing unused sl.CombinedQKVProj class * Fixing erroneous type hint * Implemnenting sl.DepthwiseConv1D in Torch * Implementing sl.MaskInvalid in Torch * Fixes for initialization * Fixes for saving weights * Removing einsums per feedback from HF staff * Removing Sequence Layers idioms from audio encoder * Fixes for reviewer comments * Converting sl.Frontend to FeatureExtractor * Updates for ConditionalGeneration.get_image_features * Adding a WIP draft of image_processing_gemma3n.py * Update modular Co-authored-by: SindhuRaghuram97 <114270661+SindhuRaghuram97@users.noreply.github.com> * Modular conversion after github suggested change * Text + image gives good results * Fixing image size preset * Draft of audio data in chat template * Removing image processing. Using SigLIP instead. * Audio input going end-to-end * Fixing dtype issues in audio encoder * x-lib formatting consistency * Adding example data * Save preprocessor_config.json from conversion script * Instrumentaiton for debugging * Additional instrumentation for preprocessing debugging * Updates to preprocessor, padding; produces correct end-to-end results on sample * Tackling configuraiton TODOs * Start of feature extractor refatcor * Adds Numpy version of USM extractor, removes Torch version and dependencies * Fixing AltUp.correct coef permute * Supporting batches of single audio segment inputs * Docstrings updates for config * In-lining audio feature extraction * Adjustments to conversion script and smoke test script --------- Co-authored-by: SindhuRaghuram97 <114270661+SindhuRaghuram97@users.noreply.github.com> Co-authored-by: Sindhu Raghuram <sindhuraghuram@google.com> Co-authored-by: pculliton <phillipculliton@gmail.com> * Gemma 3n renaming * Removing test data and utilities * Renaming test files * Gemma 3n refactor * Fix tokenizer config in conversion script * Address reviewer feedback * FeatureExtractor returns float32 by default * Adding basic tests for audio, and input name for audio encoder * Audio integration test, updates to model_id for other integration tests * Use scales for q and k norms (huggingface#26) * Update audio integration test to use HF dataset * Reviewer feedback * Expand embedding table to full vocab size in weights conversion * Mix-n-match MatFormers for Gemma 3n (huggingface#25) * Remove in-place operations (huggingface#30) * chore: removing inplace ops * remove [tensor] * n pattern * chore: reviewer feedback in AudioEncoder and AltUp * More grad clipping * Dynamo compatibility * fix: cache slicing error * chore: simplify shared kv cache slicing * chore: vision encoder rename in timm * fix: image processor do_normalize=False * fixup: style * chore: model_doc * fix: docs for code quality * chore: repo consistency * fix: RMSNorm in float as in prior Gemmas * fix: per_layer_inputs = None * chore: Gemma3nForCausalLM from Gemma3nForConditionalGeneration checkpoint * chore: repo consistency * Add initial unit tests for Gemma3nAudioFeatureExtractor (huggingface#27) * Add initial unit tests for Gemma3nAudioFeatureExtractor * Add basic unit tests for Gemma3nProcessor (huggingface#28) Co-authored-by: Douglas Reid <21148125+douglas-reid@users.noreply.github.com> * parameterize tests --------- Co-authored-by: Douglas Reid <21148125+douglas-reid@users.noreply.github.com> * chore: code style * fix: test cases * style and consistency * fix config in the test to be coherent with layer cache sharing * fix hidden states in tests and code * inits and mappings * fix modality prefixes * test order and prefixes * fix test exception * fix class order and reduce model size for faster tests * restore _checkpoint_conversion_mapping to load Caual from Conditional * fix config mapping! * fix: reviewer feedback --------- Co-authored-by: SindhuRaghuram97 <114270661+SindhuRaghuram97@users.noreply.github.com> Co-authored-by: Sindhu Raghuram <sindhuraghuram@google.com> Co-authored-by: raushan <raushan@huggingface.co> Co-authored-by: Mayank Chaturvedi <imayank@google.com> Co-authored-by: Douglas Reid <douglas-reid@users.noreply.github.com> Co-authored-by: Douglas Reid <21148125+douglas-reid@users.noreply.github.com> Co-authored-by: Xuan-Son Nguyen <thichthat@gmail.com> Co-authored-by: pculliton <phillipculliton@gmail.com> Co-authored-by: Aritra Roy Gosthipaty <aritra.born2fly@gmail.com> Co-authored-by: Cyril Vallez <cyril.vallez@gmail.com>

* testing utilities for numerics comparisons * Corrected einsum to nn.Linear weights conversion * Inherit scaled word embs from Gemma3 not Bart * Fixing transposes for collapsed linears * More transpose fixes * numpy api fix * RMSNorm: Explicit kwargs, scale_shift=0.0 when with_scale=True * Force AltUp to float32 * Updating debugging script for AudioEncoder debugging * Support divide_weight_by_sqrt_fan_in from JAX for per-layer inputs * Correcting attention einsum conversions * RMSNorm in type of x * Fixing douplicate laurel norm/gating * KV sharing using the right previous indices * Refactor kv shared index computation. Correct frac_shared_layers * Use num_shared_layers instead of inferring from a fraction * fixing a bug for logging * Fix shared data_ptrs in altup inits * rope: adjust proj -> norm -> rope to preserve computation (#20) * rope: adjust proj -> norm -> rope to preserve computation * Removing some breaking language model fluff in ConditionalGeneration * Consolidate query_states transforms --------- Co-authored-by: Douglas Reid <21148125+douglas-reid@users.noreply.github.com> Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Vectorize the loops in AltUp (#19) * Vectorize the loops in AltUp * fix typo * Expanding to support batched inputs * remove extra debug script * Fix AltUp.forward --------- Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Add 'scale_shift=0.0, with_scale=True' to the final norm in TextModel * Convert norm to 1/sqrt (#21) * Convert norm to 1/sqrt * Scale shift change per Phil's rec * Adding default activation sparsity * Fixing 2B config in weights conversion script * Fixing RMSNorm parameters - adding scale_shift and with_scale * Correcting query pre-attention scaling * Adding query_rescale_scalar to text config * Adding layer_idx to MLP * Permafix for input_layernorm * Use 1/sqrt instead of rsqrt in DecoderLayer * Fix o_proj conversion * Conversion script update for vision encoder * Removing logging for debugging timm model * Fixing bugs in Gemma3p5ForConditionalGeneration for text generation * Generating the modeling_gemma3p5.py file * Removing the addition of an erroneous line in the modeling file * Adding gemma3p5 text model to modeling_auto * Bugfix: Updating the interleaving of inputs_embeds and vision_embeds * Updating the modeling file with the latest bugfix changes * Updating models/auto for Gemma 3n * using AutoTokenizer in forward test * Adding processing_gemma3p5.py * Gemma 3n configured for AutoModel. Conversion script updated. * Removing errant merge artifacts --------- Co-authored-by: Mayank Chaturvedi <imayank@google.com> Co-authored-by: Douglas Reid <douglas-reid@users.noreply.github.com> Co-authored-by: Douglas Reid <21148125+douglas-reid@users.noreply.github.com> Co-authored-by: Xuan-Son Nguyen <thichthat@gmail.com> Co-authored-by: Sindhu Raghuram <sindhuraghuram@google.com>

* Gemma 3n * initial commit of Gemma 3n scaffold * Fixing param pass through on Gemm3p5RMSNorm * Adds Einsum layer to Gemma 3n * Updating EinsumLayer API * Undoing erroneous force push * Reverting RMSNorm to with_scale by default * Adds LAuReL to Gemma 3n * Adds AltUp to Gemma 3n * Adding Gemma3p5 overall and text config with vision and audio config placeholders (#3) * Adding gemma3p5 text configs * Adding audio config placeholders * Adding a placeholder for vision configs * Updating MobileNetVisionConfig, inheriting TimmWrapperConfig * Updating text configs * Update src/transformers/models/gemma3p5/modular_gemma3p5.py Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Removing altup configs to accept the suggested configs * Update src/transformers/models/gemma3p5/modular_gemma3p5.py Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Updating altup config * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Addressing review comments and updating text configs * Adding a config for activation sparsity * Updating configs to pass through options to super class init and adjust some name prefixes * Updating laurel and altup with corrected config values * Normalizing sub_config initializers --------- Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Updating MLP with activation sparsity (#2) * Updating DecoderBlock for Gemma 3n (#3) * Initial Gemm3nTextModel (#4) NOTE: This implementation WILL CHANGE in the coming weeks, however, changes will be strictly additive and this will remain a suitable baseline for downstream implementations to reference. * Adding KV Cache Sharing * Adds Einsum layer to Gemma 3n * Updating EinsumLayer API * Refactored kv cache sharing in attention * Adding KVStore for cache sharing * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update src/transformers/cache_utils.py Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Undoing erroneous force push * Reverting RMSNorm to with_scale by default * Adds LAuReL to Gemma 3n * Updating KV Cache Sharing implementation * Updating the q and k norm definitions in the attention module * Fixing name error for q,k,v RMS norm to use the right 3n module * Updating MLP with activation sparsity * Updating DecoderBlock for Gemma 3.5 * Updating kv cache sharing implementation with the use of a cache buffer and refactoring some lines of code * Isolating KV Cache logic to relevant components * Fixing logic error in Gemma3nAttention.forward * Refactoring caching contributions and fixing kv_store initialization * Simplifying Configs * Remove errant self from super init call * Bug fix in the Attention module - changing self.head_dim to config.head_dim * Bug fixes in the LaurelBlock and RMS Norm super init call * removing redundant code from a merge * Adding per_layer_inputs to TextModel * Adding preprocess embeddings with altup * Adds per-layer-to-single output and a host of TODOs * Integrating altup predict with the model workflow and other minor bug fixes * Using nn.Embedding temporarily for text model * It goes forward * Minor refactor of attention sparsity and RoPE initialization * Fixing duplicate rope_scaling param bug when loading from pretrained --------- Co-authored-by: Sindhu Raghuram <sindhuraghuram@google.com> Co-authored-by: SindhuRaghuram97 <114270661+SindhuRaghuram97@users.noreply.github.com> * Normalizing on altup_num_inputs config option * regenerating modeling file after syncing to HEAD * Use torch.std(..., unbiased=False) for activation sparsity (#8) * Refactoring to a single QVK Norm (#13) * AltUp: support scale_corrected_output (#14) * Converts einsums to nn.Linear (#7) * Converts einsums to nn.Linear * Removing unused variables * Aligning SharedKVCache with HybridCache (#11) * Alinging SharedKVStore with HybridCache * Remove KVStore. Refactor apply_rotary_pos_emb for sharing * Addressing review comments * Supporting split modality embeddings in Gemma3n (#10) * Adding the Embedder class * Update modular Co-authored-by: Ryan Mullins <ryan@ryanmullins.org> * Update modular Co-authored-by: Ryan Mullins <ryan@ryanmullins.org> * Update modular Co-authored-by: Ryan Mullins <ryan@ryanmullins.org> * Update modular Co-authored-by: Ryan Mullins <ryan@ryanmullins.org> * Update modular Co-authored-by: Ryan Mullins <ryan@ryanmullins.org> * Update modular Co-authored-by: Ryan Mullins <ryan@ryanmullins.org> * Addressing review comments, adding audio embedding layers, integrating embedder with the remaining architecture, adding a forward method for conditional generation * Apply suggestions from code review Co-authored-by: Ryan Mullins <ryan@ryanmullins.org> * Update modular Co-authored-by: Ryan Mullins <ryan@ryanmullins.org> * Addressing review comments, prop drilling audio and vision configs to the text config * Removing TODO's that have been addressed * Simplify Embedder init and add audio embeddings * Embeddings refactor. Adds Gemma3nAudioEmbedder and Gemma3nVisionEmbedder * Refactoring vision and audio embeddings into ConditionalGeneration model --------- Co-authored-by: Ryan Mullins <ryan@ryanmullins.org> Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Updating attention mask for Gemma 3.5 (#15) * xxx_token_index to xxx_token_id * remvoing deprecated last_cache_position * Removing references to SigLIP * Always init per-layer inputs * Using torch.finfo().min for epsilon_tensor * Gemma3nDecoderLayer inherits from Gemma3DecoderLayer. Remove gating lambdas * fix modular GEMMA3N_INPUTS_DOCSTRING * Gemma3nAttention inherits from Gemma3Attention * Modular inheritance fixes * CausalLM conversion script for 4B model (#16) * Add Gemma3n Audio Encoder (#6) * initial commit of Gemma 3.5 scaffold * Fixing param pass through on Gemm3nRMSNorm * Adds Einsum layer to Gemma 3.5 * Updating EinsumLayer API * Undoing erroneous force push * Reverting RMSNorm to with_scale by default * Adds LAuReL to Gemma 3n * Adds AltUp to Gemma 3n * Adding Gemma3n overall and text config with vision and audio config placeholders (#3) * Adding gemma3n text configs * Adding audio config placeholders * Adding a placeholder for vision configs * Updating MobileNetVisionConfig, inheriting TimmWrapperConfig * Updating text configs * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Removing altup configs to accept the suggested configs * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Updating altup config * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Addressing review comments and updating text configs * Adding a config for activation sparsity * Updating configs to pass through options to super class init and adjust some name prefixes * Updating laurel and altup with corrected config values * Normalizing sub_config initializers --------- Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Updating MLP with activation sparsity (#2) * Updating DecoderBlock for Gemma 3.5 (#3) * Initial Gemm3nTextModel (#4) NOTE: This implementation WILL CHANGE in the coming weeks, however, changes will be strictly additive and this will remain a suitable baseline for downstream implementations to reference. * Adding KV Cache Sharing * Adds Einsum layer to Gemma 3.5 * Updating EinsumLayer API * Refactored kv cache sharing in attention * Adding KVStore for cache sharing * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update src/transformers/cache_utils.py Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Undoing erroneous force push * Reverting RMSNorm to with_scale by default * Adds LAuReL to Gemma 3n * Updating KV Cache Sharing implementation * Updating the q and k norm definitions in the attention module * Fixing name error for q,k,v RMS norm to use the right Gemma 3n module * Updating MLP with activation sparsity * Updating DecoderBlock for Gemma 3.5 * Updating kv cache sharing implementation with the use of a cache buffer and refactoring some lines of code * Isolating KV Cache logic to relevant components * Fixing logic error in Gemma3nAttention.forward * Refactoring caching contributions and fixing kv_store initialization * Simplifying Configs * Remove errant self from super init call * Bug fix in the Attention module - changing self.head_dim to config.head_dim * Bug fixes in the LaurelBlock and RMS Norm super init call * removing redundant code from a merge * Adding per_layer_inputs to TextModel * Adding preprocess embeddings with altup * Adds per-layer-to-single output and a host of TODOs * Integrating altup predict with the model workflow and other minor bug fixes * Using nn.Embedding temporarily for text model * It goes forward * Minor refactor of attention sparsity and RoPE initialization * Fixing duplicate rope_scaling param bug when loading from pretrained --------- Co-authored-by: Sindhu Raghuram <sindhuraghuram@google.com> Co-authored-by: SindhuRaghuram97 <114270661+SindhuRaghuram97@users.noreply.github.com> * Normalizing on altup_num_inputs config option * Adding audio encoder config * Adds high-level components for Audio Encoder * Implement uniform reducer for Audio Encoder * Adding placeholders for Conformer components in Audio Encoder * Adding placeholders for SubSampleConvProjection components in Audio Encoder * Adding SequenceLayer component placeholders * Implementing Gemma3nAudioEncoder with nn.Sequential * Implementing Gemma3nAudioSubSampleConvProjection with nn.Sequential * Implementing Conformer model with SequenceLayers * Use OrderedDict in nn.Sequential initializers * Implements sl.Residual in Torch with nn.Sequential and OrderedDict * Adopting a base SequenceLayer class with default forward() method * Implementing sl.GatedLinearUnit in Torch * Implementing sl.Swish in Torch * Implementing sl.ReLU in Torch * Implementing sl.Scale in Torch * Removing sl.Dropout after tree-shaking * Implementing sl.RMSNorm in Torch with fake shape * Implementing sl.GroupNorm in Torch * Implementing sl.Conv2d in Torch * Implementing sl.Dense in Torch * Removing sl.Delay layers, which act as pass-throughs * Connecting shapes to configs in initializers * Removing sl.Emit * Implementing sl.ExpandDims in Torch * Adding sl.GradientClipping to Torch * Implementing sl.DenseShaped in Torch * Implementing sl.LDPA in Torch * Removing unused sl.CombinedQKVProj class * Fixing erroneous type hint * Implemnenting sl.DepthwiseConv1D in Torch * Implementing sl.MaskInvalid in Torch * Fixes for initialization * Fixes for saving weights * Removing einsums per feedback from HF staff * Removing Sequence Layers idioms from audio encoder * Fixes for reviewer comments * CausalLM conversion script for 4B model * inv_timescales to non-persistent buffer * Addressing audio encoder Attention feedback * Addressing Gemma3nAudioSSCPConvBlock feedback * Addressing Gemma3nAudioConformerAttention feedback * Addressing padding feedback * Weights conversion loads audio state dict * Always use vision_config so saving works * Token id updates for configs * Stubs for interleaving audio embs * Addressing reviewer feedback --------- Co-authored-by: SindhuRaghuram97 <114270661+SindhuRaghuram97@users.noreply.github.com> Co-authored-by: Sindhu Raghuram <sindhuraghuram@google.com> * Fixing cache access error * Removing duplicate code from a bad merge * Gemma 3n Text + Vision Part 1 (#17) * testing utilities for numerics comparisons * Corrected einsum to nn.Linear weights conversion * Inherit scaled word embs from Gemma3 not Bart * Fixing transposes for collapsed linears * More transpose fixes * numpy api fix * RMSNorm: Explicit kwargs, scale_shift=0.0 when with_scale=True * Force AltUp to float32 * Updating debugging script for AudioEncoder debugging * Support divide_weight_by_sqrt_fan_in from JAX for per-layer inputs * Correcting attention einsum conversions * RMSNorm in type of x * Fixing douplicate laurel norm/gating * KV sharing using the right previous indices * Refactor kv shared index computation. Correct frac_shared_layers * Use num_shared_layers instead of inferring from a fraction * fixing a bug for logging * Fix shared data_ptrs in altup inits * rope: adjust proj -> norm -> rope to preserve computation (#20) * rope: adjust proj -> norm -> rope to preserve computation * Removing some breaking language model fluff in ConditionalGeneration * Consolidate query_states transforms --------- Co-authored-by: Douglas Reid <21148125+douglas-reid@users.noreply.github.com> Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Vectorize the loops in AltUp (#19) * Vectorize the loops in AltUp * fix typo * Expanding to support batched inputs * remove extra debug script * Fix AltUp.forward --------- Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Add 'scale_shift=0.0, with_scale=True' to the final norm in TextModel * Convert norm to 1/sqrt (#21) * Convert norm to 1/sqrt * Scale shift change per Phil's rec * Adding default activation sparsity * Fixing 2B config in weights conversion script * Fixing RMSNorm parameters - adding scale_shift and with_scale * Correcting query pre-attention scaling * Adding query_rescale_scalar to text config * Adding layer_idx to MLP * Permafix for input_layernorm * Use 1/sqrt instead of rsqrt in DecoderLayer * Fix o_proj conversion * Conversion script update for vision encoder * Removing logging for debugging timm model * Fixing bugs in Gemma3nForConditionalGeneration for text generation * Generating the modeling_gemma3n.py file * Removing the addition of an erroneous line in the modeling file * Adding gemma3n text model to modeling_auto * Bugfix: Updating the interleaving of inputs_embeds and vision_embeds * Updating the modeling file with the latest bugfix changes * Updating models/auto for Gemma 3n * using AutoTokenizer in forward test * Adding processing_gemma3n.py * Gemma 3n configured for AutoModel. Conversion script updated. * Removing errant merge artifacts --------- Co-authored-by: Mayank Chaturvedi <imayank@google.com> Co-authored-by: Douglas Reid <douglas-reid@users.noreply.github.com> Co-authored-by: Douglas Reid <21148125+douglas-reid@users.noreply.github.com> Co-authored-by: Xuan-Son Nguyen <thichthat@gmail.com> Co-authored-by: Sindhu Raghuram <sindhuraghuram@google.com> * Removing errant debugging statements from Gemma 3 * Gemma3n audio model (#18) * testing utilities for numerics comparisons * Implement CumulativeGroupNorm and add to SubSampleConvProjection and SSCPConvBlock * Add audio version of forward script based on RyanMullins' implementation * Updating to match encoder tests. WIP: config question needs resolving * Updates to audio classes to enable end-to-end running * Removing vestigial classes, cleaning up print statements * Adding SiLU / Swish to audio conformer feed forward block * Shifted Gemma3p5Audio naming prefix to Gemma3NanoAudio * Adding outputs to audio test * Fixes to padding in SSCP and 1D convolution, align RMS Norm with wider model * Update forward test to load from local weights * Update conversion to process / output audio layers * Update __all__ to export audio encoder * AutoModel registration for Gemma 3n Audio * Use AutoModel for ConditionalGeneration.audio_tower * Fixing input_proj_linear transpose * Fixing Gemma3NanoAudioConformerAttention.post conversion * Fixing Gemma3NanoAudioSSCPConvBlock.conv weights conversion * Correcting indentation issue on Gemma3p5RMSNorm --------- Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Text + Vision Part 2 (#23) * Updates for ConditionalGeneration.get_image_features * Adding a WIP draft of image_processing_gemma3p5.py * Update src/transformers/models/gemma3p5/modular_gemma3p5.py Co-authored-by: SindhuRaghuram97 <114270661+SindhuRaghuram97@users.noreply.github.com> * Modular conversion after github suggested change * Text + image gives good results * Fixing image size preset * Updating configs for the 2B variant in the conversion script * Using final generation config in conversion script --------- Co-authored-by: Sindhu Raghuram <sindhuraghuram@google.com> Co-authored-by: SindhuRaghuram97 <114270661+SindhuRaghuram97@users.noreply.github.com> * Audio Integration (#12) * initial commit of Gemma 3n scaffold * Fixing param pass through on Gemm3nRMSNorm * Adds Einsum layer to Gemma 3n * Updating EinsumLayer API * Undoing erroneous force push * Reverting RMSNorm to with_scale by default * Adds LAuReL to Gemma 3n * Adds AltUp to Gemma 3n * Adding Gemma 3n overall and text config with vision and audio config placeholders (#3) * Adding Gemma 3n text configs * Adding audio config placeholders * Adding a placeholder for vision configs * Updating MobileNetVisionConfig, inheriting TimmWrapperConfig * Updating text configs * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Removing altup configs to accept the suggested configs * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Updating altup config * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Addressing review comments and updating text configs * Adding a config for activation sparsity * Updating configs to pass through options to super class init and adjust some name prefixes * Updating laurel and altup with corrected config values * Normalizing sub_config initializers --------- Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Updating MLP with activation sparsity (#2) * Updating DecoderBlock for Gemma 3n (#3) * Initial Gemma3nTextModel (#4) NOTE: This implementation WILL CHANGE in the coming weeks, however, changes will be strictly additive and this will remain a suitable baseline for downstream implementations to reference. * Adding KV Cache Sharing * Adds Einsum layer to Gemma 3n * Updating EinsumLayer API * Refactored kv cache sharing in attention * Adding KVStore for cache sharing * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update src/transformers/cache_utils.py Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Undoing erroneous force push * Reverting RMSNorm to with_scale by default * Adds LAuReL to Gemma 3n * Updating KV Cache Sharing implementation * Updating the q and k norm definitions in the attention module * Fixing name error for q,k,v RMS norm to use the right 3n module * Updating MLP with activation sparsity * Updating DecoderBlock for Gemma 3n * Updating kv cache sharing implementation with the use of a cache buffer and refactoring some lines of code * Isolating KV Cache logic to relevant components * Fixing logic error in Gemma3nAttention.forward * Refactoring caching contributions and fixing kv_store initialization * Simplifying Configs * Remove errant self from super init call * Bug fix in the Attention module - changing self.head_dim to config.head_dim * Bug fixes in the LaurelBlock and RMS Norm super init call * removing redundant code from a merge * Adding per_layer_inputs to TextModel * Adding preprocess embeddings with altup * Adds per-layer-to-single output and a host of TODOs * Integrating altup predict with the model workflow and other minor bug fixes * Using nn.Embedding temporarily for text model * It goes forward * Minor refactor of attention sparsity and RoPE initialization * Fixing duplicate rope_scaling param bug when loading from pretrained --------- Co-authored-by: Sindhu Raghuram <sindhuraghuram@google.com> Co-authored-by: SindhuRaghuram97 <114270661+SindhuRaghuram97@users.noreply.github.com> * Normalizing on altup_num_inputs config option * Adding audio encoder config * Adds high-level components for Audio Encoder * Implement uniform reducer for Audio Encoder * Adding placeholders for Conformer components in Audio Encoder * Adding placeholders for SubSampleConvProjection components in Audio Encoder * Adding SequenceLayer component placeholders * Implementing Gemma3nAudioEncoder with nn.Sequential * Implementing Gemma3nAudioSubSampleConvProjection with nn.Sequential * Implementing Conformer model with SequenceLayers * Use OrderedDict in nn.Sequential initializers * Implements sl.Residual in Torch with nn.Sequential and OrderedDict * Adopting a base SequenceLayer class with default forward() method * Implementing sl.GatedLinearUnit in Torch * Implementing sl.Swish in Torch * Implementing sl.ReLU in Torch * Implementing sl.Scale in Torch * Removing sl.Dropout after tree-shaking * Implementing sl.RMSNorm in Torch with fake shape * Implementing sl.GroupNorm in Torch * Implementing sl.Conv2d in Torch * Implementing sl.Dense in Torch * Removing sl.Delay layers, which act as pass-throughs * Connecting shapes to configs in initializers * Removing sl.Emit * Implementing sl.ExpandDims in Torch * Adding sl.GradientClipping to Torch * Implementing sl.DenseShaped in Torch * Implementing sl.LDPA in Torch * Removing unused sl.CombinedQKVProj class * Fixing erroneous type hint * Implemnenting sl.DepthwiseConv1D in Torch * Implementing sl.MaskInvalid in Torch * Fixes for initialization * Fixes for saving weights * Removing einsums per feedback from HF staff * Removing Sequence Layers idioms from audio encoder * Fixes for reviewer comments * Converting sl.Frontend to FeatureExtractor * Updates for ConditionalGeneration.get_image_features * Adding a WIP draft of image_processing_gemma3n.py * Update modular Co-authored-by: SindhuRaghuram97 <114270661+SindhuRaghuram97@users.noreply.github.com> * Modular conversion after github suggested change * Text + image gives good results * Fixing image size preset * Draft of audio data in chat template * Removing image processing. Using SigLIP instead. * Audio input going end-to-end * Fixing dtype issues in audio encoder * x-lib formatting consistency * Adding example data * Save preprocessor_config.json from conversion script * Instrumentaiton for debugging * Additional instrumentation for preprocessing debugging * Updates to preprocessor, padding; produces correct end-to-end results on sample * Tackling configuraiton TODOs * Start of feature extractor refatcor * Adds Numpy version of USM extractor, removes Torch version and dependencies * Fixing AltUp.correct coef permute * Supporting batches of single audio segment inputs * Docstrings updates for config * In-lining audio feature extraction * Adjustments to conversion script and smoke test script --------- Co-authored-by: SindhuRaghuram97 <114270661+SindhuRaghuram97@users.noreply.github.com> Co-authored-by: Sindhu Raghuram <sindhuraghuram@google.com> Co-authored-by: pculliton <phillipculliton@gmail.com> * Gemma 3n renaming * Removing test data and utilities * Renaming test files * Gemma 3n refactor * Fix tokenizer config in conversion script * Address reviewer feedback * FeatureExtractor returns float32 by default * Adding basic tests for audio, and input name for audio encoder * Audio integration test, updates to model_id for other integration tests * Use scales for q and k norms (#26) * Update audio integration test to use HF dataset * Reviewer feedback * Expand embedding table to full vocab size in weights conversion * Mix-n-match MatFormers for Gemma 3n (#25) * Remove in-place operations (#30) * chore: removing inplace ops * remove [tensor] * n pattern * chore: reviewer feedback in AudioEncoder and AltUp * More grad clipping * Dynamo compatibility * fix: cache slicing error * chore: simplify shared kv cache slicing * chore: vision encoder rename in timm * fix: image processor do_normalize=False * fixup: style * chore: model_doc * fix: docs for code quality * chore: repo consistency * fix: RMSNorm in float as in prior Gemmas * fix: per_layer_inputs = None * chore: Gemma3nForCausalLM from Gemma3nForConditionalGeneration checkpoint * chore: repo consistency * Add initial unit tests for Gemma3nAudioFeatureExtractor (#27) * Add initial unit tests for Gemma3nAudioFeatureExtractor * Add basic unit tests for Gemma3nProcessor (#28) Co-authored-by: Douglas Reid <21148125+douglas-reid@users.noreply.github.com> * parameterize tests --------- Co-authored-by: Douglas Reid <21148125+douglas-reid@users.noreply.github.com> * chore: code style * fix: test cases * style and consistency * fix config in the test to be coherent with layer cache sharing * fix hidden states in tests and code * inits and mappings * fix modality prefixes * test order and prefixes * fix test exception * fix class order and reduce model size for faster tests * restore _checkpoint_conversion_mapping to load Caual from Conditional * fix config mapping! * fix: reviewer feedback --------- Co-authored-by: SindhuRaghuram97 <114270661+SindhuRaghuram97@users.noreply.github.com> Co-authored-by: Sindhu Raghuram <sindhuraghuram@google.com> Co-authored-by: raushan <raushan@huggingface.co> Co-authored-by: Mayank Chaturvedi <imayank@google.com> Co-authored-by: Douglas Reid <douglas-reid@users.noreply.github.com> Co-authored-by: Douglas Reid <21148125+douglas-reid@users.noreply.github.com> Co-authored-by: Xuan-Son Nguyen <thichthat@gmail.com> Co-authored-by: pculliton <phillipculliton@gmail.com> Co-authored-by: Aritra Roy Gosthipaty <aritra.born2fly@gmail.com> Co-authored-by: Cyril Vallez <cyril.vallez@gmail.com> * fix import test * add model args * auto_docstring * replace test path * consistency * skip tests for now * fix docstring for doc builder * skip unused attr --------- Co-authored-by: SindhuRaghuram97 <114270661+SindhuRaghuram97@users.noreply.github.com> Co-authored-by: Sindhu Raghuram <sindhuraghuram@google.com> Co-authored-by: raushan <raushan@huggingface.co> Co-authored-by: Mayank Chaturvedi <imayank@google.com> Co-authored-by: Douglas Reid <douglas-reid@users.noreply.github.com> Co-authored-by: Douglas Reid <21148125+douglas-reid@users.noreply.github.com> Co-authored-by: Xuan-Son Nguyen <thichthat@gmail.com> Co-authored-by: pculliton <phillipculliton@gmail.com> Co-authored-by: Aritra Roy Gosthipaty <aritra.born2fly@gmail.com> Co-authored-by: Cyril Vallez <cyril.vallez@gmail.com> Co-authored-by: Arthur <arthur.zucker@gmail.com>

Update modelling to work with new checkpoints, exposes output_router_logits

* Gemma 3n * initial commit of Gemma 3n scaffold * Fixing param pass through on Gemm3p5RMSNorm * Adds Einsum layer to Gemma 3n * Updating EinsumLayer API * Undoing erroneous force push * Reverting RMSNorm to with_scale by default * Adds LAuReL to Gemma 3n * Adds AltUp to Gemma 3n * Adding Gemma3p5 overall and text config with vision and audio config placeholders (huggingface#3) * Adding gemma3p5 text configs * Adding audio config placeholders * Adding a placeholder for vision configs * Updating MobileNetVisionConfig, inheriting TimmWrapperConfig * Updating text configs * Update src/transformers/models/gemma3p5/modular_gemma3p5.py Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Removing altup configs to accept the suggested configs * Update src/transformers/models/gemma3p5/modular_gemma3p5.py Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Updating altup config * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Addressing review comments and updating text configs * Adding a config for activation sparsity * Updating configs to pass through options to super class init and adjust some name prefixes * Updating laurel and altup with corrected config values * Normalizing sub_config initializers --------- Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Updating MLP with activation sparsity (huggingface#2) * Updating DecoderBlock for Gemma 3n (huggingface#3) * Initial Gemm3nTextModel (huggingface#4) NOTE: This implementation WILL CHANGE in the coming weeks, however, changes will be strictly additive and this will remain a suitable baseline for downstream implementations to reference. * Adding KV Cache Sharing * Adds Einsum layer to Gemma 3n * Updating EinsumLayer API * Refactored kv cache sharing in attention * Adding KVStore for cache sharing * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update src/transformers/cache_utils.py Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Undoing erroneous force push * Reverting RMSNorm to with_scale by default * Adds LAuReL to Gemma 3n * Updating KV Cache Sharing implementation * Updating the q and k norm definitions in the attention module * Fixing name error for q,k,v RMS norm to use the right 3n module * Updating MLP with activation sparsity * Updating DecoderBlock for Gemma 3.5 * Updating kv cache sharing implementation with the use of a cache buffer and refactoring some lines of code * Isolating KV Cache logic to relevant components * Fixing logic error in Gemma3nAttention.forward * Refactoring caching contributions and fixing kv_store initialization * Simplifying Configs * Remove errant self from super init call * Bug fix in the Attention module - changing self.head_dim to config.head_dim * Bug fixes in the LaurelBlock and RMS Norm super init call * removing redundant code from a merge * Adding per_layer_inputs to TextModel * Adding preprocess embeddings with altup * Adds per-layer-to-single output and a host of TODOs * Integrating altup predict with the model workflow and other minor bug fixes * Using nn.Embedding temporarily for text model * It goes forward * Minor refactor of attention sparsity and RoPE initialization * Fixing duplicate rope_scaling param bug when loading from pretrained --------- Co-authored-by: Sindhu Raghuram <sindhuraghuram@google.com> Co-authored-by: SindhuRaghuram97 <114270661+SindhuRaghuram97@users.noreply.github.com> * Normalizing on altup_num_inputs config option * regenerating modeling file after syncing to HEAD * Use torch.std(..., unbiased=False) for activation sparsity (huggingface#8) * Refactoring to a single QVK Norm (huggingface#13) * AltUp: support scale_corrected_output (huggingface#14) * Converts einsums to nn.Linear (huggingface#7) * Converts einsums to nn.Linear * Removing unused variables * Aligning SharedKVCache with HybridCache (huggingface#11) * Alinging SharedKVStore with HybridCache * Remove KVStore. Refactor apply_rotary_pos_emb for sharing * Addressing review comments * Supporting split modality embeddings in Gemma3n (huggingface#10) * Adding the Embedder class * Update modular Co-authored-by: Ryan Mullins <ryan@ryanmullins.org> * Update modular Co-authored-by: Ryan Mullins <ryan@ryanmullins.org> * Update modular Co-authored-by: Ryan Mullins <ryan@ryanmullins.org> * Update modular Co-authored-by: Ryan Mullins <ryan@ryanmullins.org> * Update modular Co-authored-by: Ryan Mullins <ryan@ryanmullins.org> * Update modular Co-authored-by: Ryan Mullins <ryan@ryanmullins.org> * Addressing review comments, adding audio embedding layers, integrating embedder with the remaining architecture, adding a forward method for conditional generation * Apply suggestions from code review Co-authored-by: Ryan Mullins <ryan@ryanmullins.org> * Update modular Co-authored-by: Ryan Mullins <ryan@ryanmullins.org> * Addressing review comments, prop drilling audio and vision configs to the text config * Removing TODO's that have been addressed * Simplify Embedder init and add audio embeddings * Embeddings refactor. Adds Gemma3nAudioEmbedder and Gemma3nVisionEmbedder * Refactoring vision and audio embeddings into ConditionalGeneration model --------- Co-authored-by: Ryan Mullins <ryan@ryanmullins.org> Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Updating attention mask for Gemma 3.5 (huggingface#15) * xxx_token_index to xxx_token_id * remvoing deprecated last_cache_position * Removing references to SigLIP * Always init per-layer inputs * Using torch.finfo().min for epsilon_tensor * Gemma3nDecoderLayer inherits from Gemma3DecoderLayer. Remove gating lambdas * fix modular GEMMA3N_INPUTS_DOCSTRING * Gemma3nAttention inherits from Gemma3Attention * Modular inheritance fixes * CausalLM conversion script for 4B model (huggingface#16) * Add Gemma3n Audio Encoder (huggingface#6) * initial commit of Gemma 3.5 scaffold * Fixing param pass through on Gemm3nRMSNorm * Adds Einsum layer to Gemma 3.5 * Updating EinsumLayer API * Undoing erroneous force push * Reverting RMSNorm to with_scale by default * Adds LAuReL to Gemma 3n * Adds AltUp to Gemma 3n * Adding Gemma3n overall and text config with vision and audio config placeholders (huggingface#3) * Adding gemma3n text configs * Adding audio config placeholders * Adding a placeholder for vision configs * Updating MobileNetVisionConfig, inheriting TimmWrapperConfig * Updating text configs * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Removing altup configs to accept the suggested configs * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Updating altup config * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Addressing review comments and updating text configs * Adding a config for activation sparsity * Updating configs to pass through options to super class init and adjust some name prefixes * Updating laurel and altup with corrected config values * Normalizing sub_config initializers --------- Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Updating MLP with activation sparsity (huggingface#2) * Updating DecoderBlock for Gemma 3.5 (huggingface#3) * Initial Gemm3nTextModel (huggingface#4) NOTE: This implementation WILL CHANGE in the coming weeks, however, changes will be strictly additive and this will remain a suitable baseline for downstream implementations to reference. * Adding KV Cache Sharing * Adds Einsum layer to Gemma 3.5 * Updating EinsumLayer API * Refactored kv cache sharing in attention * Adding KVStore for cache sharing * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update src/transformers/cache_utils.py Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Undoing erroneous force push * Reverting RMSNorm to with_scale by default * Adds LAuReL to Gemma 3n * Updating KV Cache Sharing implementation * Updating the q and k norm definitions in the attention module * Fixing name error for q,k,v RMS norm to use the right Gemma 3n module * Updating MLP with activation sparsity * Updating DecoderBlock for Gemma 3.5 * Updating kv cache sharing implementation with the use of a cache buffer and refactoring some lines of code * Isolating KV Cache logic to relevant components * Fixing logic error in Gemma3nAttention.forward * Refactoring caching contributions and fixing kv_store initialization * Simplifying Configs * Remove errant self from super init call * Bug fix in the Attention module - changing self.head_dim to config.head_dim * Bug fixes in the LaurelBlock and RMS Norm super init call * removing redundant code from a merge * Adding per_layer_inputs to TextModel * Adding preprocess embeddings with altup * Adds per-layer-to-single output and a host of TODOs * Integrating altup predict with the model workflow and other minor bug fixes * Using nn.Embedding temporarily for text model * It goes forward * Minor refactor of attention sparsity and RoPE initialization * Fixing duplicate rope_scaling param bug when loading from pretrained --------- Co-authored-by: Sindhu Raghuram <sindhuraghuram@google.com> Co-authored-by: SindhuRaghuram97 <114270661+SindhuRaghuram97@users.noreply.github.com> * Normalizing on altup_num_inputs config option * Adding audio encoder config * Adds high-level components for Audio Encoder * Implement uniform reducer for Audio Encoder * Adding placeholders for Conformer components in Audio Encoder * Adding placeholders for SubSampleConvProjection components in Audio Encoder * Adding SequenceLayer component placeholders * Implementing Gemma3nAudioEncoder with nn.Sequential * Implementing Gemma3nAudioSubSampleConvProjection with nn.Sequential * Implementing Conformer model with SequenceLayers * Use OrderedDict in nn.Sequential initializers * Implements sl.Residual in Torch with nn.Sequential and OrderedDict * Adopting a base SequenceLayer class with default forward() method * Implementing sl.GatedLinearUnit in Torch * Implementing sl.Swish in Torch * Implementing sl.ReLU in Torch * Implementing sl.Scale in Torch * Removing sl.Dropout after tree-shaking * Implementing sl.RMSNorm in Torch with fake shape * Implementing sl.GroupNorm in Torch * Implementing sl.Conv2d in Torch * Implementing sl.Dense in Torch * Removing sl.Delay layers, which act as pass-throughs * Connecting shapes to configs in initializers * Removing sl.Emit * Implementing sl.ExpandDims in Torch * Adding sl.GradientClipping to Torch * Implementing sl.DenseShaped in Torch * Implementing sl.LDPA in Torch * Removing unused sl.CombinedQKVProj class * Fixing erroneous type hint * Implemnenting sl.DepthwiseConv1D in Torch * Implementing sl.MaskInvalid in Torch * Fixes for initialization * Fixes for saving weights * Removing einsums per feedback from HF staff * Removing Sequence Layers idioms from audio encoder * Fixes for reviewer comments * CausalLM conversion script for 4B model * inv_timescales to non-persistent buffer * Addressing audio encoder Attention feedback * Addressing Gemma3nAudioSSCPConvBlock feedback * Addressing Gemma3nAudioConformerAttention feedback * Addressing padding feedback * Weights conversion loads audio state dict * Always use vision_config so saving works * Token id updates for configs * Stubs for interleaving audio embs * Addressing reviewer feedback --------- Co-authored-by: SindhuRaghuram97 <114270661+SindhuRaghuram97@users.noreply.github.com> Co-authored-by: Sindhu Raghuram <sindhuraghuram@google.com> * Fixing cache access error * Removing duplicate code from a bad merge * Gemma 3n Text + Vision Part 1 (huggingface#17) * testing utilities for numerics comparisons * Corrected einsum to nn.Linear weights conversion * Inherit scaled word embs from Gemma3 not Bart * Fixing transposes for collapsed linears * More transpose fixes * numpy api fix * RMSNorm: Explicit kwargs, scale_shift=0.0 when with_scale=True * Force AltUp to float32 * Updating debugging script for AudioEncoder debugging * Support divide_weight_by_sqrt_fan_in from JAX for per-layer inputs * Correcting attention einsum conversions * RMSNorm in type of x * Fixing douplicate laurel norm/gating * KV sharing using the right previous indices * Refactor kv shared index computation. Correct frac_shared_layers * Use num_shared_layers instead of inferring from a fraction * fixing a bug for logging * Fix shared data_ptrs in altup inits * rope: adjust proj -> norm -> rope to preserve computation (huggingface#20) * rope: adjust proj -> norm -> rope to preserve computation * Removing some breaking language model fluff in ConditionalGeneration * Consolidate query_states transforms --------- Co-authored-by: Douglas Reid <21148125+douglas-reid@users.noreply.github.com> Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Vectorize the loops in AltUp (huggingface#19) * Vectorize the loops in AltUp * fix typo * Expanding to support batched inputs * remove extra debug script * Fix AltUp.forward --------- Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Add 'scale_shift=0.0, with_scale=True' to the final norm in TextModel * Convert norm to 1/sqrt (huggingface#21) * Convert norm to 1/sqrt * Scale shift change per Phil's rec * Adding default activation sparsity * Fixing 2B config in weights conversion script * Fixing RMSNorm parameters - adding scale_shift and with_scale * Correcting query pre-attention scaling * Adding query_rescale_scalar to text config * Adding layer_idx to MLP * Permafix for input_layernorm * Use 1/sqrt instead of rsqrt in DecoderLayer * Fix o_proj conversion * Conversion script update for vision encoder * Removing logging for debugging timm model * Fixing bugs in Gemma3nForConditionalGeneration for text generation * Generating the modeling_gemma3n.py file * Removing the addition of an erroneous line in the modeling file * Adding gemma3n text model to modeling_auto * Bugfix: Updating the interleaving of inputs_embeds and vision_embeds * Updating the modeling file with the latest bugfix changes * Updating models/auto for Gemma 3n * using AutoTokenizer in forward test * Adding processing_gemma3n.py * Gemma 3n configured for AutoModel. Conversion script updated. * Removing errant merge artifacts --------- Co-authored-by: Mayank Chaturvedi <imayank@google.com> Co-authored-by: Douglas Reid <douglas-reid@users.noreply.github.com> Co-authored-by: Douglas Reid <21148125+douglas-reid@users.noreply.github.com> Co-authored-by: Xuan-Son Nguyen <thichthat@gmail.com> Co-authored-by: Sindhu Raghuram <sindhuraghuram@google.com> * Removing errant debugging statements from Gemma 3 * Gemma3n audio model (huggingface#18) * testing utilities for numerics comparisons * Implement CumulativeGroupNorm and add to SubSampleConvProjection and SSCPConvBlock * Add audio version of forward script based on RyanMullins' implementation * Updating to match encoder tests. WIP: config question needs resolving * Updates to audio classes to enable end-to-end running * Removing vestigial classes, cleaning up print statements * Adding SiLU / Swish to audio conformer feed forward block * Shifted Gemma3p5Audio naming prefix to Gemma3NanoAudio * Adding outputs to audio test * Fixes to padding in SSCP and 1D convolution, align RMS Norm with wider model * Update forward test to load from local weights * Update conversion to process / output audio layers * Update __all__ to export audio encoder * AutoModel registration for Gemma 3n Audio * Use AutoModel for ConditionalGeneration.audio_tower * Fixing input_proj_linear transpose * Fixing Gemma3NanoAudioConformerAttention.post conversion * Fixing Gemma3NanoAudioSSCPConvBlock.conv weights conversion * Correcting indentation issue on Gemma3p5RMSNorm --------- Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Text + Vision Part 2 (huggingface#23) * Updates for ConditionalGeneration.get_image_features * Adding a WIP draft of image_processing_gemma3p5.py * Update src/transformers/models/gemma3p5/modular_gemma3p5.py Co-authored-by: SindhuRaghuram97 <114270661+SindhuRaghuram97@users.noreply.github.com> * Modular conversion after github suggested change * Text + image gives good results * Fixing image size preset * Updating configs for the 2B variant in the conversion script * Using final generation config in conversion script --------- Co-authored-by: Sindhu Raghuram <sindhuraghuram@google.com> Co-authored-by: SindhuRaghuram97 <114270661+SindhuRaghuram97@users.noreply.github.com> * Audio Integration (huggingface#12) * initial commit of Gemma 3n scaffold * Fixing param pass through on Gemm3nRMSNorm * Adds Einsum layer to Gemma 3n * Updating EinsumLayer API * Undoing erroneous force push * Reverting RMSNorm to with_scale by default * Adds LAuReL to Gemma 3n * Adds AltUp to Gemma 3n * Adding Gemma 3n overall and text config with vision and audio config placeholders (huggingface#3) * Adding Gemma 3n text configs * Adding audio config placeholders * Adding a placeholder for vision configs * Updating MobileNetVisionConfig, inheriting TimmWrapperConfig * Updating text configs * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Removing altup configs to accept the suggested configs * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Updating altup config * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Addressing review comments and updating text configs * Adding a config for activation sparsity * Updating configs to pass through options to super class init and adjust some name prefixes * Updating laurel and altup with corrected config values * Normalizing sub_config initializers --------- Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Updating MLP with activation sparsity (huggingface#2) * Updating DecoderBlock for Gemma 3n (huggingface#3) * Initial Gemma3nTextModel (huggingface#4) NOTE: This implementation WILL CHANGE in the coming weeks, however, changes will be strictly additive and this will remain a suitable baseline for downstream implementations to reference. * Adding KV Cache Sharing * Adds Einsum layer to Gemma 3n * Updating EinsumLayer API * Refactored kv cache sharing in attention * Adding KVStore for cache sharing * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update modular Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Update src/transformers/cache_utils.py Co-authored-by: Ryan Mullins <ryanmullins@google.com> * Undoing erroneous force push * Reverting RMSNorm to with_scale by default * Adds LAuReL to Gemma 3n * Updating KV Cache Sharing implementation * Updating the q and k norm definitions in the attention module * Fixing name error for q,k,v RMS norm to use the right 3n module * Updating MLP with activation sparsity * Updating DecoderBlock for Gemma 3n * Updating kv cache sharing implementation with the use of a cache buffer and refactoring some lines of code * Isolating KV Cache logic to relevant components * Fixing logic error in Gemma3nAttention.forward * Refactoring caching contributions and fixing kv_store initialization * Simplifying Configs * Remove errant self from super init call * Bug fix in the Attention module - changing self.head_dim to config.head_dim * Bug fixes in the LaurelBlock and RMS Norm super init call * removing redundant code from a merge * Adding per_layer_inputs to TextModel * Adding preprocess embeddings with altup * Adds per-layer-to-single output and a host of TODOs * Integrating altup predict with the model workflow and other minor bug fixes * Using nn.Embedding temporarily for text model * It goes forward * Minor refactor of attention sparsity and RoPE initialization * Fixing duplicate rope_scaling param bug when loading from pretrained --------- Co-authored-by: Sindhu Raghuram <sindhuraghuram@google.com> Co-authored-by: SindhuRaghuram97 <114270661+SindhuRaghuram97@users.noreply.github.com> * Normalizing on altup_num_inputs config option * Adding audio encoder config * Adds high-level components for Audio Encoder * Implement uniform reducer for Audio Encoder * Adding placeholders for Conformer components in Audio Encoder * Adding placeholders for SubSampleConvProjection components in Audio Encoder * Adding SequenceLayer component placeholders * Implementing Gemma3nAudioEncoder with nn.Sequential * Implementing Gemma3nAudioSubSampleConvProjection with nn.Sequential * Implementing Conformer model with SequenceLayers * Use OrderedDict in nn.Sequential initializers * Implements sl.Residual in Torch with nn.Sequential and OrderedDict * Adopting a base SequenceLayer class with default forward() method * Implementing sl.GatedLinearUnit in Torch * Implementing sl.Swish in Torch * Implementing sl.ReLU in Torch * Implementing sl.Scale in Torch * Removing sl.Dropout after tree-shaking * Implementing sl.RMSNorm in Torch with fake shape * Implementing sl.GroupNorm in Torch * Implementing sl.Conv2d in Torch * Implementing sl.Dense in Torch * Removing sl.Delay layers, which act as pass-throughs * Connecting shapes to configs in initializers * Removing sl.Emit * Implementing sl.ExpandDims in Torch * Adding sl.GradientClipping to Torch * Implementing sl.DenseShaped in Torch * Implementing sl.LDPA in Torch * Removing unused sl.CombinedQKVProj class * Fixing erroneous type hint * Implemnenting sl.DepthwiseConv1D in Torch * Implementing sl.MaskInvalid in Torch * Fixes for initialization * Fixes for saving weights * Removing einsums per feedback from HF staff * Removing Sequence Layers idioms from audio encoder * Fixes for reviewer comments * Converting sl.Frontend to FeatureExtractor * Updates for ConditionalGeneration.get_image_features * Adding a WIP draft of image_processing_gemma3n.py * Update modular Co-authored-by: SindhuRaghuram97 <114270661+SindhuRaghuram97@users.noreply.github.com> * Modular conversion after github suggested change * Text + image gives good results * Fixing image size preset * Draft of audio data in chat template * Removing image processing. Using SigLIP instead. * Audio input going end-to-end * Fixing dtype issues in audio encoder * x-lib formatting consistency * Adding example data * Save preprocessor_config.json from conversion script * Instrumentaiton for debugging * Additional instrumentation for preprocessing debugging * Updates to preprocessor, padding; produces correct end-to-end results on sample * Tackling configuraiton TODOs * Start of feature extractor refatcor * Adds Numpy version of USM extractor, removes Torch version and dependencies * Fixing AltUp.correct coef permute * Supporting batches of single audio segment inputs * Docstrings updates for config * In-lining audio feature extraction * Adjustments to conversion script and smoke test script --------- Co-authored-by: SindhuRaghuram97 <114270661+SindhuRaghuram97@users.noreply.github.com> Co-authored-by: Sindhu Raghuram <sindhuraghuram@google.com> Co-authored-by: pculliton <phillipculliton@gmail.com> * Gemma 3n renaming * Removing test data and utilities * Renaming test files * Gemma 3n refactor * Fix tokenizer config in conversion script * Address reviewer feedback * FeatureExtractor returns float32 by default * Adding basic tests for audio, and input name for audio encoder * Audio integration test, updates to model_id for other integration tests * Use scales for q and k norms (huggingface#26) * Update audio integration test to use HF dataset * Reviewer feedback * Expand embedding table to full vocab size in weights conversion * Mix-n-match MatFormers for Gemma 3n (huggingface#25) * Remove in-place operations (huggingface#30) * chore: removing inplace ops * remove [tensor] * n pattern * chore: reviewer feedback in AudioEncoder and AltUp * More grad clipping * Dynamo compatibility * fix: cache slicing error * chore: simplify shared kv cache slicing * chore: vision encoder rename in timm * fix: image processor do_normalize=False * fixup: style * chore: model_doc * fix: docs for code quality * chore: repo consistency * fix: RMSNorm in float as in prior Gemmas * fix: per_layer_inputs = None * chore: Gemma3nForCausalLM from Gemma3nForConditionalGeneration checkpoint * chore: repo consistency * Add initial unit tests for Gemma3nAudioFeatureExtractor (huggingface#27) * Add initial unit tests for Gemma3nAudioFeatureExtractor * Add basic unit tests for Gemma3nProcessor (huggingface#28) Co-authored-by: Douglas Reid <21148125+douglas-reid@users.noreply.github.com> * parameterize tests --------- Co-authored-by: Douglas Reid <21148125+douglas-reid@users.noreply.github.com> * chore: code style * fix: test cases * style and consistency * fix config in the test to be coherent with layer cache sharing * fix hidden states in tests and code * inits and mappings * fix modality prefixes * test order and prefixes * fix test exception * fix class order and reduce model size for faster tests * restore _checkpoint_conversion_mapping to load Caual from Conditional * fix config mapping! * fix: reviewer feedback --------- Co-authored-by: SindhuRaghuram97 <114270661+SindhuRaghuram97@users.noreply.github.com> Co-authored-by: Sindhu Raghuram <sindhuraghuram@google.com> Co-authored-by: raushan <raushan@huggingface.co> Co-authored-by: Mayank Chaturvedi <imayank@google.com> Co-authored-by: Douglas Reid <douglas-reid@users.noreply.github.com> Co-authored-by: Douglas Reid <21148125+douglas-reid@users.noreply.github.com> Co-authored-by: Xuan-Son Nguyen <thichthat@gmail.com> Co-authored-by: pculliton <phillipculliton@gmail.com> Co-authored-by: Aritra Roy Gosthipaty <aritra.born2fly@gmail.com> Co-authored-by: Cyril Vallez <cyril.vallez@gmail.com> * fix import test * add model args * auto_docstring * replace test path * consistency * skip tests for now * fix docstring for doc builder * skip unused attr --------- Co-authored-by: SindhuRaghuram97 <114270661+SindhuRaghuram97@users.noreply.github.com> Co-authored-by: Sindhu Raghuram <sindhuraghuram@google.com> Co-authored-by: raushan <raushan@huggingface.co> Co-authored-by: Mayank Chaturvedi <imayank@google.com> Co-authored-by: Douglas Reid <douglas-reid@users.noreply.github.com> Co-authored-by: Douglas Reid <21148125+douglas-reid@users.noreply.github.com> Co-authored-by: Xuan-Son Nguyen <thichthat@gmail.com> Co-authored-by: pculliton <phillipculliton@gmail.com> Co-authored-by: Aritra Roy Gosthipaty <aritra.born2fly@gmail.com> Co-authored-by: Cyril Vallez <cyril.vallez@gmail.com> Co-authored-by: Arthur <arthur.zucker@gmail.com>

Fix some glitches in extract_features.py

451a3e7

1. Add no_cuda argument 2. Fix the "not all arguments converted during string formatting" error thrown at line 230

thomwolf closed this Nov 17, 2018

maeotaku mentioned this pull request May 23, 2019

bert->onnx ->caffe2 weird error #633

Closed

HongyanJiao mentioned this pull request Sep 19, 2019

traced_model #1291

Closed

manchandasahil mentioned this pull request Mar 22, 2021

Longformer training : CUDA error: device-side assert triggered #10852

Closed

2 tasks

jambayk pushed a commit to jambayk/transformers that referenced this pull request Apr 7, 2022

Merge pull request huggingface#21 from microsoft/aibhanda/dev_options

e4e57f0

Add ORT internal options config flag

jameshennessytempus pushed a commit to jameshennessytempus/transformers that referenced this pull request Jun 1, 2023

Merge pull request huggingface#21 from jamesthesnake/main

dc329c8

j

lwmlyy mentioned this pull request Aug 15, 2023

add util for ram efficient loading of model when using fsdp #25107

Merged

1 task

ocavue pushed a commit to ocavue/transformers that referenced this pull request Sep 13, 2023

Merge pull request huggingface#21 from xenova/whisper

9c3c29d

[whisper] Add support for audio clips > 30 seconds

SangbumChoi added a commit to SangbumChoi/transformers that referenced this pull request Sep 24, 2024

Merge pull request huggingface#21 from Superb-AI-Suite/process_semantics

49f5036

expand the first dimension of input_semantics to batch size

jmamou added a commit to jmamou/transformers that referenced this pull request Feb 27, 2025

Fix Joao's comments (huggingface#21)

751a099

* remove threading * fix logits_processor * fix test device

ArthurZucker pushed a commit that referenced this pull request Apr 5, 2025

Merge pull request #21 from huggingface/add_fbgemm

aa8daba

Adding fbgemm

RyanMullins mentioned this pull request Jun 26, 2025

Gemma 3n #39059

Merged

5 tasks

ArthurZucker added a commit that referenced this pull request Aug 5, 2025

Merge pull request #21 from huggingface/ed-fix-modeling

410435a

Update modelling to work with new checkpoints, exposes output_router_logits

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Fix some glitches in extract_features.py #21

Fix some glitches in extract_features.py #21

Uh oh!

cnrpman commented Nov 15, 2018

thomwolf commented Nov 17, 2018

Labels

2 participants

Fix some glitches in extract_features.py #21

Fix some glitches in extract_features.py #21

Uh oh!

Conversation

cnrpman commented Nov 15, 2018

thomwolf commented Nov 17, 2018

Labels

2 participants