Add Phi-3.5-vision by Dahlbomii · Pull Request #36036 · huggingface/transformers

Dahlbomii · 2025-02-04T22:20:42Z

Draft PR for now, still need to add tests and convert to modular

cc @Rocketknight1

Fixes #36071

Rocketknight1 · 2025-02-05T17:19:10Z

Looks good so far! The TODO list from here is:

Make the transformers imports relative imports so that we stop getting circular import errors in the CI
Add the phi3_5.md doc (you can copy the layout from a similar model and copy the text from the model card)
pip install transformers[quality] and make fixup or make style to get the repo consistency checks green

Once everything is green, then:

Add tests (you can copy them from another similar VLM and just change model names etc.)

Once the new tests are green as well, then:

Convert the modeling and configuration files to modular, then regenerating the modeling/config files and confirm everything passes

At that point, the PR should be finished!

ArthurZucker · 2025-02-13T15:40:39Z

Fixes #36166!

Dahlbomii · 2025-02-24T00:21:41Z

I've run into an issue when trying to run the model. From Modeling_phi3_v.py I'm getting "AttributeError: 'DynamicCache' object has no attribute 'get_max_length'" This might be due to a change to the DynamicCache class, but I really can't tell!

zucchini-nlp · 2025-02-24T08:39:50Z

@Dahlbomii hey! Yes, we deprecated get_max_length from cache a few versions ago, now it has to be Cache.get_max_cache_shape(layer_idx)

Rocketknight1 · 2025-02-24T17:18:48Z

Woah - this means the remote code version of Phi-3.5-vision-instruct is broken on main right now and requires a fixed older version of transformers. Definitely makes this PR more urgent/important!

Rocketknight1 · 2025-02-28T15:48:17Z

cc @zucchini-nlp @gante I can't figure this one out either! The test_generate_continue_from_inputs_embeds test is failing for Phi-3V but I don't really understand how generate() handles inputs_embeds - when I added some breakpoints it seems to generate an input_ids tensor with shape (batch_size, 0), which then causes errors in Phi.

zucchini-nlp · 2025-02-28T15:56:22Z

Seems like Phi-3.5V was not updated for the latest changes we had in generation, probably it has its own custom generation loop. For this test, we can call super().prepare_input_for_generation(kwargs) to get all the inputs processed correctly. Then image related kwargs can be set to None if needed. You can take a look on how we did it in other VLMs, hope it solves the issue

Rocketknight1 · 2025-02-28T16:18:54Z

@zucchini-nlp yes, that was it, sorry! They don't override generate() but they do override prepare_input_for_generation() and discard inputs_embeds after the first step.

Dahlbomii · 2025-03-14T08:49:20Z

@Rocketknight1 for reasons that escape me, when the test processor tries to call self.get_component for the tokenizer, it breaks. Any insight as to why?

Dahlbomii · 2025-04-01T09:10:58Z

@Rocketknight1 ALRIGHT I got most of the tests into the green, but I have no idea what I'm doing with the chat template stuff!

Rocketknight1 · 2025-04-01T14:51:36Z

test_apply_chat_template() isn't really necessary - I think it just copied over from smolvlm! You can remove it.

Dahlbomii · 2025-04-07T21:46:40Z

@Rocketknight1 With that I think it might be ready for review

Rocketknight1

Congrats on the PR! This looks almost ready, and the modular bit looks like it was really annoying. There are a lot of classes that look like they should be inheritable from somewhere else in the codebase, but they're just different enough that you can't.

cc @zucchini-nlp I made some comments, but you're more familiar with VLMs than me, can you review and see if anything else should be changed before we ping a core maintainer?

Rocketknight1 · 2025-04-08T16:04:49Z

+class Phi3RMSNorm(nn.Module):
+    def __init__(self, hidden_size, eps=1e-6):
+        """
+        Phi3RMSNorm is equivalent to T5LayerNorm
+        """
+        super().__init__()
+        self.weight = nn.Parameter(torch.ones(hidden_size))
+        self.variance_epsilon = eps
+
+    def forward(self, hidden_states):
+        input_dtype = hidden_states.dtype
+        hidden_states = hidden_states.to(torch.float32)
+        variance = hidden_states.pow(2).mean(-1, keepdim=True)
+        hidden_states = hidden_states * torch.rsqrt(variance + self.variance_epsilon)
+        return self.weight * hidden_states.to(input_dtype)


I think this can just be inherited from a class like T5LayerNorm!

Rocketknight1 · 2025-04-08T16:06:36Z

+class Phi3RotaryEmbedding(nn.Module):
+    def __init__(self, dim, max_position_embeddings=2048, base=10000, device=None):
+        super().__init__()
+
+        self.dim = dim
+        self.max_position_embeddings = max_position_embeddings
+        self.base = base
+        self.register_buffer("inv_freq", None, persistent=False)
+
+    @torch.no_grad()
+    def forward(self, x, position_ids, seq_len=None):
+        # x: [bs, num_attention_heads, seq_len, head_size]
+        if self.inv_freq is None:
+            self.inv_freq = 1.0 / (
+                self.base ** (torch.arange(0, self.dim, 2, dtype=torch.int64, device=x.device).float() / self.dim)
+            )
+        inv_freq_expanded = self.inv_freq[None, :, None].float().expand(position_ids.shape[0], -1, 1)
+        position_ids_expanded = position_ids[:, None, :].float()
+        # Force float32 since bfloat16 loses precision on long contexts
+        # See https://github.com/huggingface/transformers/pull/29285
+        device_type = x.device.type
+        device_type = device_type if isinstance(device_type, str) and device_type != "mps" else "cpu"
+        with torch.autocast(device_type=device_type, enabled=False):
+            freqs = (inv_freq_expanded.float() @ position_ids_expanded.float()).transpose(1, 2)
+            emb = torch.cat((freqs, freqs), dim=-1)
+            cos = emb.cos()
+            sin = emb.sin()
+        return cos.to(dtype=x.dtype), sin.to(dtype=x.dtype)


Although the code is a little different, I think this could be inherited from another RotaryEmbedding class in the library, with the same outputs. Maybe open_llama?

I think it is pretty much same as phi3 or phi3-MoE. We also support rope scaling and dynamic rope with unified API, adding a decorator will do the thing

Rocketknight1 · 2025-04-08T16:08:06Z

+class Phi3MLP(nn.Module):
+    def __init__(self, config):
+        super().__init__()
+
+        self.config = config
+        self.gate_up_proj = nn.Linear(config.hidden_size, 2 * config.intermediate_size, bias=False)
+        self.down_proj = nn.Linear(config.intermediate_size, config.hidden_size, bias=False)
+
+        self.activation_fn = ACT2FN[config.hidden_act]
+
+    def forward(self, hidden_states: torch.FloatTensor) -> torch.FloatTensor:
+        up_states = self.gate_up_proj(hidden_states)
+
+        gate, up_states = up_states.chunk(2, dim=-1)
+        up_states = up_states * self.activation_fn(gate)
+
+        return self.down_proj(up_states)


I think this could be inherited from Phi3!

Rocketknight1 · 2025-04-08T16:09:17Z

+        return self.down_proj(up_states)
+
+
+class Phi3Attention(nn.Module):


Although the code is a little different, I think you could maybe inherit the layer from Phi3 and not need this code here! (But you'd have to test to be sure)

+1, seems pretty much same. The qkv can be split in conversion script if needed and we can inherit from Phi3, which also adds new attention interface (easier TGI, vLLM integrations)

Rocketknight1 · 2025-04-08T16:11:22Z

+}
+
+
+class Phi3DecoderLayer(nn.Module):


Although the code is a little different, I think this could inherit from the equivalent class in Phi3 without output changes (maybe!)

Rocketknight1 · 2025-04-08T16:12:44Z

+    _supports_sdpa = False
+    _supports_cache_class = True
+
+    _version = "0.0.5"


Suggested change

_version = "0.0.5"

Probably unnecessary!

zucchini-nlp

@Dahlbomii thanks a lot for working on this!

The PR seems to be adapted from custom code in the hub mostly, and I realize that custom code is outdated and doesn't follow transformers standards. Would be nice to do little bit more clean up before merging, by unifying text and vision backbones as AutoModel (I guess it is identical to CLIP and Phi3) and leave only the Base/ConditionalLM multimodal class in modular

I left a few comments below on how we can do that. LMK if that makes sense

zucchini-nlp · 2025-04-08T16:31:29Z

@@ -0,0 +1,59 @@
+<!--Copyright 2024 The HuggingFace Team. All rights reserved.


zucchini-nlp · 2025-04-08T16:32:09Z

+
+
+


Let's add the abstract and finalize docs before pinging core maintainer

zucchini-nlp · 2025-04-08T16:33:25Z

        ("persimmon", "PersimmonForCausalLM"),
        ("phi", "PhiForCausalLM"),
        ("phi3", "Phi3ForCausalLM"),
+        ("phi3_v", "Phi3VForCausalLM"),


has to be in ImageTextToText mapping when it works with "image+text". AutoCausalLM is currently reserved for text modality only

zucchini-nlp · 2025-04-08T16:44:15Z

+class Phi3RotaryEmbedding(nn.Module):
+    def __init__(self, dim, max_position_embeddings=2048, base=10000, device=None):
+        super().__init__()
+
+        self.dim = dim
+        self.max_position_embeddings = max_position_embeddings
+        self.base = base
+        self.register_buffer("inv_freq", None, persistent=False)
+
+    @torch.no_grad()
+    def forward(self, x, position_ids, seq_len=None):
+        # x: [bs, num_attention_heads, seq_len, head_size]
+        if self.inv_freq is None:
+            self.inv_freq = 1.0 / (
+                self.base ** (torch.arange(0, self.dim, 2, dtype=torch.int64, device=x.device).float() / self.dim)
+            )
+        inv_freq_expanded = self.inv_freq[None, :, None].float().expand(position_ids.shape[0], -1, 1)
+        position_ids_expanded = position_ids[:, None, :].float()
+        # Force float32 since bfloat16 loses precision on long contexts
+        # See https://github.com/huggingface/transformers/pull/29285
+        device_type = x.device.type
+        device_type = device_type if isinstance(device_type, str) and device_type != "mps" else "cpu"
+        with torch.autocast(device_type=device_type, enabled=False):
+            freqs = (inv_freq_expanded.float() @ position_ids_expanded.float()).transpose(1, 2)
+            emb = torch.cat((freqs, freqs), dim=-1)
+            cos = emb.cos()
+            sin = emb.sin()
+        return cos.to(dtype=x.dtype), sin.to(dtype=x.dtype)


I think it is pretty much same as phi3 or phi3-MoE. We also support rope scaling and dynamic rope with unified API, adding a decorator will do the thing

zucchini-nlp · 2025-04-08T16:45:29Z

+    def _init_rope(self):
+        if self.rope_scaling is None:
+            self.rotary_emb = Phi3RotaryEmbedding(
+                self.head_dim,
+                max_position_embeddings=self.max_position_embeddings,
+                base=self.rope_theta,
+            )
+        else:
+            scaling_type = self.config.rope_scaling["type"]
+            if scaling_type == "su":
+                self.rotary_emb = Phi3SuScaledRotaryEmbedding(self.head_dim, self.config)
+            elif scaling_type == "yarn":
+                self.rotary_emb = Phi3YarnScaledRotaryEmbedding(self.head_dim, self.config)
+            else:
+                raise ValueError(f"Unknown RoPE scaling type {scaling_type}")


we init RoPE once per model in base class, this is an old way and has to be removed

zucchini-nlp · 2025-04-08T17:10:01Z

+        images: ImageInput = None,
+        padding: Union[bool, str, PaddingStrategy] = False,
+        truncation: Union[bool, str, TruncationStrategy] = None,
+        max_length=None,
+        return_tensors: Optional[Union[str, TensorType]] = TensorType.PYTORCH,
+        add_special_tokens: bool = True,
+    ) -> BatchFeature:


let's add kwargs with standard processing API. For ex in llava

transformers/src/transformers/models/llava_next/processing_llava_next.py

Lines 32 to 41 in 08e3217

class LlavaNextProcessorKwargs(ProcessingKwargs, total=False):

_defaults = {

"text_kwargs": {

"padding": False,

},

"images_kwargs": {

"do_pad": True,

},

}

zucchini-nlp · 2025-04-08T17:11:09Z

+    def calc_num_image_tokens(self, images: ImageInput):
+        """Calculate the number of image tokens for each image.
+        Args:
+            images (`ImageInput`):
+                Image to preprocess. Expects a single or batch of images with pixel values ranging from 0 to 255. If
+                passing in images with pixel values between 0 and 1, set `do_rescale=False`.
+        """
+        return self.image_processor.calc_num_image_tokens(images)
+
+    def calc_num_image_tokens_from_image_size(self, width, height):
+        """Calculate the number of image token for an image with given width and height.
+        Args:
+            width (`int`):
+                Width of the image.
+            height (`int`):
+                Height of the image.
+        """
+        return self.image_processor.calc_num_image_tokens_from_image_size(width, height)
+


can we move these all from image processor to processor?

zucchini-nlp · 2025-04-08T17:15:48Z

+        input_ids = torch.tensor(input_ids, dtype=torch.long).unsqueeze(0)
+        attention_mask = (input_ids > -1000000).to(torch.long)


inputs are converted to tensor only if asked by users, so we have to put it in Batchfeature and let it handle all type conversion

zucchini-nlp · 2025-04-08T17:19:31Z

+
+        pattern = r"<\|image_\d+\|>"
+        prompt_chunks = [self.tokenizer(chunk).input_ids for chunk in re.split(pattern, texts)]
+
+        if "num_img_tokens" in images:
+            num_img_tokens = images["num_img_tokens"]
+        else:
+            assert "num_crops" in images, "num_crops must be provided in images if num_img_tokens is not provided"
+            num_crops = images["num_crops"]
+            num_img_tokens = [_num_crops * self.num_img_tokens for _num_crops in num_crops]
+
+        images, image_sizes = images["pixel_values"], images["image_sizes"]
+
+        # image_tags needs to start from 1 to n
+        image_tags = re.findall(pattern, texts)
+        # image_ids = [int(s.split("|")[1].split("_")[-1]) * -1 for s in image_tags]
+        # image_ids_pad = [[iid]*num_img_tokens[i] for i, iid in enumerate(image_ids)]
+        image_ids = [int(s.split("|")[1].split("_")[-1]) for s in image_tags]
+        unique_image_ids = sorted(set(image_ids))
+        # image_ids must start from 1, and must be continuous int, e.g. [1, 2, 3], cannot be [1, 4, 5]
+        # check the condition
+        assert unique_image_ids == list(range(1, len(unique_image_ids) + 1)), (
+            f"image_ids must start from 1, and must be continuous int, e.g. [1, 2, 3], cannot be {unique_image_ids}"
+        )
+        # total images must be the same as the number of image tags
+        assert len(unique_image_ids) == len(images), (
+            f"total images must be the same as the number of image tags, got {len(unique_image_ids)} image tags and {len(images)} images"


this looks too complicated! From what I see adding special number on image tokens is not needed, since what happens here is

text = "<|image|> What is this?" text_after_separator = " <placeholder> <placeholder> [<placeholder> .... ] What is this?""

This is same as in many other VLMs, and the simplest case is LLaVA. Let's clean up a bit and make nececssary changes to processor config when needed

zucchini-nlp · 2025-04-08T17:21:02Z

+            token_labels,
+            choice_labels,
+        ) = config_and_inputs
+        inputs_dict = {"input_ids": input_ids, "attention_mask": input_mask}


pixel_values have to be in inputs, so we test same way as model will be used by users

ArthurZucker · 2025-07-07T12:16:25Z

cc @Dahlbomii removing my review until @zucchini-nlp's comments are adressed!

MiNeves00 · 2025-07-28T11:23:02Z

Any news on this? Phi-3.5-vision has been out for a while, just wanted to know an ETA if there is one

zucchini-nlp · 2025-08-01T09:17:02Z

I will try to finish this PR next week

Dahlbomii · 2025-08-15T22:48:40Z

I'm sorry, I left this on the back burner for far longer than I thought! @zucchini-nlp I don't know if you've already started working on this, but I could pick it back up now if it's still up in the air! Whatever works best, really.

Rocketknight1 · 2025-08-20T12:43:33Z

            ("paligemma", ("SiglipImageProcessor", "SiglipImageProcessorFast")),
            ("perceiver", ("PerceiverImageProcessor", "PerceiverImageProcessorFast")),
            ("perception_lm", (None, "PerceptionLMImageProcessorFast")),
+            ("phi3_v", ("Phi3VImageProcessor",)),


Suggested change

("phi3_v", ("Phi3VImageProcessor",)),

("phi3_v", ("Phi3VImageProcessor", None)),

I think a lot of the test failures are caused by this line here! We changed the format of this list, so every model needs an ImageProcessor and an ImageProcessorFast, which can be None.

github-actions · 2025-08-20T18:54:15Z

[For maintainers] Suggested jobs to run (before merge)

run-slow: auto, phi3_v

Dahlbomii · 2025-08-21T22:12:18Z

@zucchini-nlp I could use a little help; I've been trying to get the model running again for testing, but another PR removed get_usable_length from cache_utils and I'm unsure how to adjust the model for this change. If it helps I tracked down the removal, #39106

yaswanth19 · 2025-08-22T10:06:46Z

Hey @Dahlbomii I can see that a lot of modules need standardizing and refactoring to transformers API. I can help with the PR to push changes quickly with my previous experience so that the model can be merged faster. LMK if you’re open to that and if so please add me as a collaborator to this PR.

zucchini-nlp · 2025-09-08T13:23:47Z

@Dahlbomii any progress on the PR? We would love to get it merged soon as it will help us to support the new Phi-Grounding model. If not, I'm happy to take over or ask @yaswanth19 for help if they have bandwidth

yaswanth19 · 2025-09-08T14:41:16Z

@zucchini-nlp Since there isn't much activity here for some time. I will start working on this and raise a new PR 🫡

Dahlbomii · 2025-09-08T21:55:23Z

I've been working on getting things back in line with codebase changes, but I still hadn't finished enough to push any commits! I'm sure @yaswanth19 could handle things faster though, I can make you a collaborator still if that helps!

I can also share some of what I've been looking through: phi3_v's SuScaledRotaryEmbedding class is perfectly in line with the rope update's 'longrope' type, so a change to the config and using phi3's RotaryEmbedding class works out fine there. However I couldn't figure out how to preserve phi3_v's YarnScaledRotaryEmbedding, as it still uses the long and short rope factors. Maybe your experience can help with that class, I've been at a loss.

yaswanth19 · 2025-09-09T00:08:31Z

Thanks @Dahlbomii for the confirmation, AFAIK the the text backbone is phi3 language model itself so you don't have to worry about rope - simply utilize phi3 model or utilize modular.

I can make you a collaborator still if that helps!

Thanks, although I will open a new PR as this one is a bit clumsy and is in middle ground b/w custom code and ideal code.

1himan · 2026-01-08T10:50:02Z

@Rocketknight1, should close this as duplicated of #41977

Rocketknight1 · 2026-01-12T17:04:02Z

Closing!

Dahlbomii force-pushed the add_phi3-5_vision branch from 4911ef9 to 7e16056 Compare February 11, 2025 23:02

Rocketknight1 mentioned this pull request Feb 13, 2025

Add phi3 Vision #36166

Open

2 tasks

Dahlbomii force-pushed the add_phi3-5_vision branch from 1f31710 to 0848b01 Compare March 3, 2025 23:06

Rocketknight1 mentioned this pull request Mar 10, 2025

modeling_phi3 errors with AttributeError: 'DynamicCache' object has no attribute 'get_max_length' #36071

Closed

4 tasks

Rocketknight1 marked this pull request as ready for review March 10, 2025 19:09

github-actions Bot requested review from ArthurZucker and Rocketknight1 March 10, 2025 19:09

Rocketknight1 force-pushed the add_phi3-5_vision branch from 52f6b48 to 56c8b9f Compare March 10, 2025 19:11

Dahlbomii force-pushed the add_phi3-5_vision branch from 404d2a7 to f4684f7 Compare April 2, 2025 21:48

Rocketknight1 force-pushed the add_phi3-5_vision branch from 08109e0 to c26b108 Compare April 7, 2025 17:35

Rocketknight1 reviewed Apr 8, 2025

View reviewed changes

zucchini-nlp reviewed Apr 9, 2025

View reviewed changes

ArthurZucker added the New model label Jul 7, 2025

ArthurZucker removed their request for review July 7, 2025 12:16

Dahlbomii and others added 16 commits August 19, 2025 15:14

Small adjustments to test processor

9bbfb06

Stop trying to use a slow processor bc it's cursed

eaace73

test processor updates and slight changes to processing

e5e2649

removed unnecessary tests

85aab70

maked fix-copies

5ba3fb9

maked fix-copies

9e01018

Modular file added

dcb8e27

Minor adjustments to modular and test processor

60f85e6

Minor adjustment to modular it's FINE

c2a939b

removed vestigial import

72b0140

More import changes

5e61ff2

missing variable

d7f4b42

various fixes

ea3ddd1

import adjustments

96a3d77

_flash_supports_window_size only in flashattn contexts

54e7c9a

Move things around to stop annoying modular

c78a141

Dahlbomii force-pushed the add_phi3-5_vision branch from c26b108 to c78a141 Compare August 19, 2025 22:25

Rocketknight1 reviewed Aug 20, 2025

View reviewed changes

small tuple fix

2cb360b

yaswanth19 mentioned this pull request Nov 2, 2025

Add Phi3.5 Vision Model #41977

Open

Rocketknight1 closed this Jan 12, 2026

		return self.down_proj(up_states)


		class Phi3Attention(nn.Module):

		@@ -0,0 +1,59 @@
		<!--Copyright 2024 The HuggingFace Team. All rights reserved.

	class LlavaNextProcessorKwargs(ProcessingKwargs, total=False):
	_defaults = {
	"text_kwargs": {
	"padding": False,
	},
	"images_kwargs": {
	"do_pad": True,
	},
	}

		input_ids = torch.tensor(input_ids, dtype=torch.long).unsqueeze(0)
		attention_mask = (input_ids > -1000000).to(torch.long)

	("phi3_v", ("Phi3VImageProcessor",)),
	("phi3_v", ("Phi3VImageProcessor", None)),

Conversation

Dahlbomii commented Feb 4, 2025 • edited by Rocketknight1 Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Rocketknight1 commented Feb 5, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

ArthurZucker commented Feb 13, 2025

Uh oh!

Dahlbomii commented Feb 24, 2025

Uh oh!

zucchini-nlp commented Feb 24, 2025

Uh oh!

Rocketknight1 commented Feb 24, 2025

Uh oh!

Rocketknight1 commented Feb 28, 2025

Uh oh!

zucchini-nlp commented Feb 28, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Rocketknight1 commented Feb 28, 2025

Uh oh!

Dahlbomii commented Mar 14, 2025

Uh oh!

Dahlbomii commented Apr 1, 2025

Uh oh!

Rocketknight1 commented Apr 1, 2025

Uh oh!

Dahlbomii commented Apr 7, 2025

Uh oh!

Rocketknight1 left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

zucchini-nlp left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ArthurZucker commented Jul 7, 2025

Uh oh!

MiNeves00 commented Jul 28, 2025

Uh oh!

zucchini-nlp commented Aug 1, 2025

Uh oh!

Dahlbomii commented Aug 15, 2025

Dahlbomii commented Feb 4, 2025 •

edited by Rocketknight1

Loading

Rocketknight1 commented Feb 5, 2025 •

edited

Loading

zucchini-nlp commented Feb 28, 2025 •

edited

Loading

zucchini-nlp left a comment •

edited

Loading