feat(server): support RefinedWeb models

server/text_generation_server/models/__init__.py

@@ -10,6 +10,7 @@ from text_generation_server.models.causal_lm import CausalLM
 from text_generation_server.models.flash_causal_lm import FlashCausalLM
 from text_generation_server.models.bloom import BLOOM, BLOOMSharded
 from text_generation_server.models.seq2seq_lm import Seq2SeqLM
+from text_generation_server.models.rw import RW
 from text_generation_server.models.opt import OPT, OPTSharded
 from text_generation_server.models.galactica import Galactica, GalacticaSharded
 from text_generation_server.models.santacoder import SantaCoder
@@ -30,6 +31,7 @@ try:
             )
 
         from text_generation_server.models.flash_neox import FlashNeoX, FlashNeoXSharded
+        from text_generation_server.models.flash_rw import FlashRW
         from text_generation_server.models.flash_llama import (
             FlashLlama,
             FlashLlamaSharded,
@@ -68,6 +70,7 @@ __all__ = [
 if FLASH_ATTENTION:
     __all__.append(FlashNeoX)
     __all__.append(FlashNeoXSharded)
+    __all__.append(FlashRW)
     __all__.append(FlashSantacoder)
     __all__.append(FlashSantacoderSharded)
     __all__.append(FlashLlama)
@@ -194,6 +197,34 @@ def get_model(
                 trust_remote_code=trust_remote_code,
             )
 
+    if model_type in ["RefinedWeb", "RefinedWebModel"]:
+        if sharded:
+            if FLASH_ATTENTION:
+                if config.alibi:
+                    raise NotImplementedError("sharded is not supported for this model")
+                # return FlashRWSharded(
+                #     model_id,
+                #     revision,
+                #     quantize=quantize,
+                #     trust_remote_code=trust_remote_code,
+                # )
+            raise NotImplementedError(FLASH_ATT_ERROR_MESSAGE.format(f"Sharded RefinedWeb"))
+        else:
+            if FLASH_ATTENTION and not config.alibi:
+                return FlashRW(
+                    model_id,
+                    revision,
+                    quantize=quantize,
+                    trust_remote_code=trust_remote_code,
+                )
+            else:
+                return RW(
+                    model_id,
+                    revision,
+                    quantize=quantize,
+                    trust_remote_code=trust_remote_code,
+                )
+
     if model_type == "llama":
         if sharded:
             if FLASH_ATTENTION:

server/text_generation_server/models/custom_modeling/flash_llama_modeling.py

@@ -134,20 +134,23 @@ class FlashLlamaAttention(torch.nn.Module):
     ):
         qkv = self.query_key_value(hidden_states)
         qkv = qkv.view(-1, 3, self.num_heads, self.head_size)
-        qkv_rot = self.rotary_emb(qkv, cos, sin)
+
+        # Inplace rotary
+        self.rotary_emb(qkv[:, 0], cos, sin)
+        self.rotary_emb(qkv[:, 1], cos, sin)
 
         # Prefill
         if layer_past_present_indices is None:
             # Copy to layer past
-            layer_past[...] = qkv_rot[:, 1:]
+            layer_past[...] = qkv[:, 1:]
 
             # output
-            attn_output = torch.empty_like(qkv_rot[:, 0])
+            attn_output = torch.empty_like(qkv[:, 0])
             # flash attention
             flash_attn_cuda.fwd(
-                qkv_rot[:, 0],
+                qkv[:, 0],
-                qkv_rot[:, 1],
+                qkv[:, 1],
-                qkv_rot[:, 2],
+                qkv[:, 2],
                 attn_output,
                 cu_seqlens,
                 cu_seqlens,
@@ -163,9 +166,9 @@ class FlashLlamaAttention(torch.nn.Module):
             )
         # Decode
         else:
-            query = qkv_rot[:, 0]
+            query = qkv[:, 0]
             # Add present to the layer_past tensor at the correct indices
-            layer_past[layer_past_present_indices] = qkv_rot[:, 1:]
+            layer_past[layer_past_present_indices] = qkv[:, 1:]
 
             # output
             attn_output = torch.empty_like(query)

server/text_generation_server/models/custom_modeling/flash_neox_modeling.py

@@ -101,20 +101,23 @@ class FlashNeoxAttention(torch.nn.Module):
     ):
         qkv = self.query_key_value(hidden_states)
         qkv = qkv.view(-1, 3, self.num_heads, self.head_size)
-        qkv_rot = self.rotary_emb(qkv, cos, sin)
+
+        # Inplace rotary
+        self.rotary_emb(qkv[:, 0], cos, sin)
+        self.rotary_emb(qkv[:, 1], cos, sin)
 
         # Prefill
         if layer_past_present_indices is None:
             # Copy to layer past
-            layer_past[...] = qkv_rot[:, 1:]
+            layer_past[...] = qkv[:, 1:]
 
             # output
-            attn_output = torch.empty_like(qkv_rot[:, 0])
+            attn_output = torch.empty_like(qkv[:, 0])
             # flash attention
             flash_attn_cuda.fwd(
-                qkv_rot[:, 0],
+                qkv[:, 0],
-                qkv_rot[:, 1],
+                qkv[:, 1],
-                qkv_rot[:, 2],
+                qkv[:, 2],
                 attn_output,
                 cu_seqlens,
                 cu_seqlens,
@@ -130,9 +133,9 @@ class FlashNeoxAttention(torch.nn.Module):
             )
         # Decode
         else:
-            query = qkv_rot[:, 0]
+            query = qkv[:, 0]
             # Add present to the layer_past tensor at the correct indices
-            layer_past[layer_past_present_indices] = qkv_rot[:, 1:]
+            layer_past[layer_past_present_indices] = qkv[:, 1:]
 
             # output
             attn_output = torch.empty_like(query)

server/text_generation_server/models/custom_modeling/flash_rw_modeling.py

@@ -0,0 +1,507 @@
+import torch
+import torch.distributed
+
+from loguru import logger
+from torch import nn
+from transformers.modeling_utils import PreTrainedModel
+from transformers.configuration_utils import PretrainedConfig
+from typing import Optional
+
+# Flash attention imports
+import flash_attn_cuda
+
+from text_generation_server.utils.layers import (
+    FastLinear,
+    TensorParallelRowLinear,
+    TensorParallelColumnLinear,
+    TensorParallelEmbedding,
+    FastLayerNorm,
+    PositionRotaryEmbedding,
+)
+
+
+class RWConfig(PretrainedConfig):
+    attribute_map = {
+        "num_hidden_layers": "n_layer",
+        "num_attention_heads": "n_head",
+    }
+
+    def __init__(
+            self,
+            model_type="RefinedWeb",
+            vocab_size=250880,
+            hidden_size=64,
+            n_layer=2,
+            n_head=8,
+            layer_norm_epsilon=1e-5,
+            initializer_range=0.02,
+            use_cache=True,
+            bos_token_id=1,
+            eos_token_id=2,
+            hidden_dropout=0.0,
+            attention_dropout=0.0,
+            n_head_kv=None,
+            multi_query=False,
+            alibi=False,
+            bias=False,
+            parallel_attn=False,
+            **kwargs,
+    ):
+        if alibi:
+            raise NotImplementedError("alibi is not supported by this version of the model")
+
+        self.model_type = model_type
+        self.alibi = False
+        self.rotary = True
+
+        self.vocab_size = vocab_size
+        # Backward compatibility with n_embed kwarg
+        n_embed = kwargs.pop("n_embed", None)
+        self.hidden_size = hidden_size if n_embed is None else n_embed
+        self.n_layer = n_layer
+        self.n_head = n_head
+        self.layer_norm_epsilon = layer_norm_epsilon
+        self.initializer_range = initializer_range
+        self.use_cache = use_cache
+        self.hidden_dropout = hidden_dropout
+        self.attention_dropout = attention_dropout
+        self.bias = bias
+        self.parallel_attn = parallel_attn
+
+        self.bos_token_id = bos_token_id
+        self.eos_token_id = eos_token_id
+
+        if n_head_kv is not None:
+            self.n_head_kv = n_head_kv
+        else:
+            self.n_head_kv = 1 if multi_query else n_head
+
+        super().__init__(bos_token_id=bos_token_id, eos_token_id=eos_token_id, **kwargs)
+
+
+class FlashRWAttention(torch.nn.Module):
+    def __init__(
+            self,
+            num_heads,
+            num_heads_kv,
+            hidden_size,
+            bias,
+            process_group=None,
+            reduce=True,
+    ):
+        super().__init__()
+        self.num_heads = num_heads
+        self.num_heads_kv = num_heads_kv
+        self.hidden_size = hidden_size
+        self.head_size = hidden_size // num_heads
+
+        self.rotary_emb = PositionRotaryEmbedding(self.head_size, base=10000)
+        self.softmax_scale = self.head_size ** (-0.5)
+
+        if process_group is None:
+            self.query_key_value = FastLinear(hidden_size, self.head_size * (self.num_heads + 2 * self.num_heads_kv),
+                                              bias=bias)
+            self.dense = FastLinear(hidden_size, hidden_size, bias=bias)
+        else:
+            self.num_heads = self.num_heads // process_group.size()
+            self.query_key_value = FastLinear(hidden_size, self.head_size * (self.num_heads + 2 * self.num_heads_kv),
+                                              bias=bias)
+            self.dense = TensorParallelRowLinear(
+                hidden_size, hidden_size, bias=bias, process_group=process_group, reduce=reduce
+            )
+
+    def forward(
+            self,
+            hidden_states,
+            cos,
+            sin,
+            cu_seqlens,
+            max_s,
+            layer_past,
+            layer_past_present_indices,
+            cu_seqlens_q,
+    ):
+        qkv = self.query_key_value(hidden_states)
+
+        # Split query from key_value
+        query, kv = qkv.split(
+            [self.head_size * self.num_heads, 2 * self.head_size], dim=1
+        )
+
+        # Prepare query and key_value for indexing
+        query = query.view(-1, self.num_heads, self.head_size)
+        kv = kv.view(-1, 2, 1, self.head_size)
+
+        # Inplace rotary
+        self.rotary_emb(query, cos, sin)
+        self.rotary_emb(kv[:, 0], cos, sin)
+
+        # Prefill
+        if layer_past_present_indices is None:
+            # Copy to layer past
+            layer_past[...] = kv
+            # Expand to query shape
+            kv = kv.expand(-1, 2, query.shape[1], self.head_size)
+
+            # output
+            attn_output = torch.empty_like(query)
+            # flash attention
+            flash_attn_cuda.fwd(
+                query,
+                kv[:, 0],
+                kv[:, 1],
+                attn_output,
+                cu_seqlens,
+                cu_seqlens,
+                max_s,
+                max_s,
+                0.0,
+                self.softmax_scale,
+                False,
+                True,
+                False,
+                0,
+                None,
+            )
+        # Decode
+        else:
+            # Add present to the layer_past tensor at the correct indices
+            layer_past[layer_past_present_indices] = kv
+            # Expand to query shape
+            kv = layer_past.expand(-1, 2, query.shape[1], self.head_size)
+
+            # output
+            attn_output = torch.empty_like(query)
+            # flash attention
+            flash_attn_cuda.fwd(
+                query,
+                kv[:, 0],
+                kv[:, 1],
+                attn_output,
+                cu_seqlens_q,
+                cu_seqlens,
+                1,
+                max_s,
+                0.0,
+                self.softmax_scale,
+                False,
+                False,
+                False,
+                0,
+                None,
+            )
+
+        return self.dense(attn_output.view(-1, self.num_heads * self.head_size))
+
+
+class FlashMLP(nn.Module):
+    def __init__(
+            self, hidden_size, bias, process_group=None, reduce=True
+    ):
+        super().__init__()
+        self.act = torch.nn.functional.gelu
+
+        if process_group is None:
+            self.dense_h_to_4h = FastLinear(hidden_size, 4 * hidden_size, bias=bias)
+            self.dense_4h_to_h = FastLinear(4 * hidden_size, hidden_size, bias=bias)
+        else:
+            self.dense_h_to_4h = TensorParallelColumnLinear(
+                hidden_size,
+                4 * hidden_size, bias=bias,
+                process_group=process_group,
+            )
+            self.dense_4h_to_h = TensorParallelRowLinear(
+                4 * hidden_size,
+                hidden_size, bias=bias,
+                process_group=process_group,
+                reduce=reduce,
+            )
+        self.process_group = process_group
+
+    def forward(self, hidden_states):
+        hidden_states = self.dense_h_to_4h(hidden_states)
+        hidden_states = self.act(hidden_states)
+        hidden_states = self.dense_4h_to_h(hidden_states)
+        return hidden_states
+
+
+class FlashRWLayer(nn.Module):
+    def __init__(
+            self,
+            num_heads,
+            num_heads_kv,
+            hidden_size,
+            layer_norm_eps,
+            parallel_attn,
+            process_group=None,
+    ):
+        super().__init__()
+
+        self.parallel_attn = parallel_attn
+
+        self.input_layernorm = FastLayerNorm(hidden_size, eps=layer_norm_eps)
+        self.self_attention = FlashRWAttention(num_heads, num_heads_kv, hidden_size, process_group, reduce=False)
+        self.post_attention_layernorm = FastLayerNorm(hidden_size, eps=layer_norm_eps) if not parallel_attn else None
+
+        self.mlp = FlashMLP(hidden_size, process_group, reduce=False)
+
+        self.process_group = process_group
+
+    def forward(
+            self,
+            hidden_states,
+            residual,
+            cos,
+            sin,
+            cu_seqlens,
+            max_s,
+            layer_past,
+            layer_past_present_indices,
+            cu_seqlens_q,
+    ):
+        if self.parallel_attn:
+            ln_hidden_states, residual = self.input_layernorm(hidden_states, residual)
+
+            attn_output = self.self_attention(
+                ln_hidden_states,
+                cos,
+                sin,
+                cu_seqlens,
+                max_s,
+                layer_past,
+                layer_past_present_indices,
+                cu_seqlens_q,
+            )
+
+            mlp_output = self.mlp(ln_hidden_states)
+            intermediate = mlp_output + attn_output
+
+            # Only reduce once and after the addition instead of once per layer
+            if self.process_group is not None:
+                torch.distributed.all_reduce(intermediate, group=self.process_group)
+
+            return intermediate, residual
+        else:
+            hidden_states, residual = self.input_layernorm(hidden_states, residual)
+
+            hidden_states = self.self_attention(
+                hidden_states,
+                cos,
+                sin,
+                cu_seqlens,
+                max_s,
+                layer_past,
+                layer_past_present_indices,
+                cu_seqlens_q,
+            )
+
+            hidden_states, residual = self.post_attention_layernorm(
+                hidden_states, residual
+            )
+
+            mlp_output = self.mlp(hidden_states)
+
+            return mlp_output, residual
+
+class FlashRWPreTrainedModel(PreTrainedModel):
+    config_class = RWConfig
+    supports_gradient_checkpointing = False
+    _no_split_modules = None
+
+
+class FlashRWModel(FlashRWPreTrainedModel):
+    def __init__(self, config, process_group=None):
+        super().__init__(config)
+        self.config = config
+
+        self.tp_embeddings = False
+        if process_group is not None:
+            self.tp_rank = process_group.rank()
+            self.tp_world_size = process_group.size()
+            if config.vocab_size % self.tp_world_size == 0:
+                self.tp_embeddings = True
+
+        if self.tp_embeddings:
+            self.word_embeddings = TensorParallelEmbedding(
+                config.vocab_size, config.hidden_size, process_group=process_group
+            )
+        else:
+            self.word_embeddings = nn.Embedding(config.vocab_size, config.hidden_size)
+
+        self.h = nn.ModuleList(
+            [
+                FlashRWLayer(
+                    config.n_head,
+                    config.n_head_kv,
+                    config.hidden_size,
+                    config.layer_norm_epsilon,
+                    config.parallel_attn,
+                    process_group,
+                )
+                for _ in range(config.num_hidden_layers)
+            ]
+        )
+        self.ln_f = FastLayerNorm(
+            config.hidden_size, eps=config.layer_norm_epsilon
+        )
+
+        self.gradient_checkpointing = False
+
+        self.head_size = self.h[0].self_attention.head_size
+        self.num_heads_kv = self.h[0].self_attention.num_heads_kv
+
+    def post_load_weights(self, quantize: Optional[str] = None):
+        if isinstance(self.word_embeddings, TensorParallelEmbedding):
+            self.word_embeddings.add_null_idx()
+        for layer in self.h:
+            layer: FlashRWLayer
+            layer.self_attention.query_key_value.prepare_weights(quantize)
+            layer.self_attention.dense.prepare_weights(quantize)
+            layer.mlp.dense_h_to_4h.prepare_weights(quantize)
+            layer.mlp.dense_4h_to_h.prepare_weights(quantize)
+
+    @classmethod
+    def from_pretrained(cls, pretrained_model_name_or_path, *model_args, **kwargs):
+        # Pop here as we will replace the layer in our own logic and don't want from_pretrained
+        # to do it for us
+        load_in_8bit = kwargs.pop("load_in_8bit", False)
+        model = super(FlashRWModel, cls).from_pretrained(
+            pretrained_model_name_or_path, load_in_8bit=False, *model_args, **kwargs
+        )
+
+        model.post_load_weights("bitsandbytes" if load_in_8bit else None)
+        return model
+
+    def forward(
+            self,
+            input_ids,
+            position_ids,
+            cu_seqlens,
+            cu_seqlens_q,
+            max_s,
+            past_key_values=None,
+            pre_allocate_past_size: Optional[int] = None,
+    ):
+        hidden_states = self.word_embeddings(input_ids)
+
+        # Prefill
+        if past_key_values is None:
+            # Create past tensor
+            past_key_values = hidden_states.new_empty(
+                (
+                    len(self.h),
+                    len(hidden_states)
+                    if pre_allocate_past_size is None
+                    else pre_allocate_past_size,
+                    2,
+                    self.num_heads_kv,
+                    self.head_size,
+                )
+            )
+            layer_past_present_indices = None
+            slice_past_index = len(hidden_states)
+        # Decode
+        else:
+            # Create indices from cumulative sequence lengths
+            layer_past_present_indices = cu_seqlens[1:] - 1
+            slice_past_index = None
+
+        # Get rotary cos and sin for this forward
+        # Avoid to index in each layer
+        cos, sin = self.h[0].self_attention.rotary_emb.get_cos_sin(
+            position_ids, max_s, hidden_states.dtype
+        )
+
+        residual = None
+        for i, layer in enumerate(self.h):
+            # We added padding that we now need to slice
+            layer_past_key_values = (
+                past_key_values[i]
+                if slice_past_index is None
+                else past_key_values[i, :slice_past_index]
+            )
+
+            hidden_states, residual = layer(
+                hidden_states,
+                residual,
+                cos,
+                sin,
+                cu_seqlens,
+                max_s,
+                layer_past_key_values,
+                layer_past_present_indices,
+                cu_seqlens_q,
+            )
+
+        hidden_states, _ = self.ln_f(hidden_states, residual)
+
+        return hidden_states, past_key_values
+
+
+class FlashRWForCausalLM(FlashRWPreTrainedModel):
+    def __init__(self, config, process_group=None):
+        super().__init__(config)
+
+        self.process_group = process_group
+        if self.process_group is not None:
+            self.world_size = self.process_group.size()
+        else:
+            self.world_size = 1
+
+        self.transformer = FlashRWModel(config, process_group)
+
+        if self.transformer.tp_embeddings:
+            self.lm_head = FastLinear(
+                config.hidden_size,
+                config.vocab_size // process_group.size(),
+                bias=False,
+            )
+        else:
+            self.lm_head = FastLinear(
+                config.hidden_size, config.vocab_size, bias=False
+            )
+
+    def post_load_weights(self, quantize: Optional[str] = None):
+        self.transformer.post_load_weights(quantize)
+        self.lm_head.prepare_weights()
+
+    @classmethod
+    def from_pretrained(cls, pretrained_model_name_or_path, *model_args, **kwargs):
+        # Pop here as we will replace the layer in our own logic and don't want from_pretrained
+        # to do it for us
+        load_in_8bit = kwargs.pop("load_in_8bit", False)
+        model = super(FlashRWForCausalLM, cls).from_pretrained(
+            pretrained_model_name_or_path, load_in_8bit=False, *model_args, **kwargs
+        )
+        model.post_load_weights("bitsandbytes" if load_in_8bit else None)
+        return model
+
+    def forward(
+            self,
+            input_ids,
+            position_ids,
+            cu_seqlens,
+            cu_seqlens_q,
+            max_s,
+            past_key_values: Optional[torch.Tensor] = None,
+            pre_allocate_past_size: Optional[int] = None,
+    ):
+        hidden_states, present = self.transformer(
+            input_ids,
+            position_ids,
+            cu_seqlens,
+            cu_seqlens_q,
+            max_s,
+            past_key_values,
+            pre_allocate_past_size,
+        )
+        logits = self.lm_head(hidden_states)
+
+        if self.transformer.tp_embeddings:
+            # Logits are sharded, so we need to gather them
+            world_logits = [torch.empty_like(logits) for _ in range(self.world_size)]
+            torch.distributed.all_gather(world_logits, logits, group=self.process_group)
+            world_logits = torch.cat(world_logits, dim=1)
+
+            return world_logits, present
+        return logits, present

server/text_generation_server/models/flash_rw.py

@@ -0,0 +1,246 @@
+import torch
+import torch.distributed
+
+from pathlib import Path
+from accelerate import init_empty_weights
+from opentelemetry import trace
+from safetensors import safe_open
+from transformers import AutoTokenizer, AutoConfig
+from typing import Optional, List
+
+from text_generation_server.models import FlashCausalLM
+from text_generation_server.models.custom_modeling.flash_rw_modeling import (
+    RWConfig,
+    FlashRWForCausalLM,
+    TensorParallelEmbedding,
+    TensorParallelRowLinear,
+    TensorParallelColumnLinear,
+)
+from text_generation_server.utils import (
+    initialize_torch_distributed,
+    weight_files,
+    download_weights,
+    weight_hub_files,
+    LocalEntryNotFoundError,
+)
+
+tracer = trace.get_tracer(__name__)
+
+
+class FlashRW(FlashCausalLM):
+    def __init__(
+        self,
+        model_id: str,
+        revision: Optional[str] = None,
+        quantize: Optional[str] = None,
+        trust_remote_code: bool = False,
+    ):
+        if torch.cuda.is_available():
+            device = torch.device("cuda")
+            dtype = torch.float16
+        else:
+            raise NotImplementedError("RW is only available on GPU")
+
+        tokenizer = AutoTokenizer.from_pretrained(
+            model_id,
+            revision=revision,
+            padding_side="left",
+            truncation_side="left",
+            trust_remote_code=trust_remote_code,
+        )
+
+        config = RWConfig.from_pretrained(
+            model_id,
+            revision=revision,
+        )
+        from loguru import logger
+        logger.error(config.model_type)
+
+        # We do not use from_pretrained as we modified the model internal module layout
+        try:
+            filenames = weight_files(model_id, revision, ".bin")
+        # Local files not found
+        except LocalEntryNotFoundError:
+            hub_files = weight_hub_files(model_id, revision, ".bin")
+            filenames = download_weights(hub_files, model_id, revision)
+
+        with init_empty_weights():
+            model = FlashRWForCausalLM(config)
+
+        self.load_weights(
+            model,
+            filenames,
+            quantize,
+            device,
+            dtype,
+        )
+
+        super(FlashCausalLM, self).__init__(
+            model=model.to(device),
+            tokenizer=tokenizer,
+            requires_padding=False,
+            dtype=dtype,
+            device=device,
+        )
+
+    @staticmethod
+    def load_weights(
+        model: FlashRWForCausalLM,
+        filenames: List[Path],
+        quantize: Optional[str],
+        device: torch.device,
+        dtype: torch.dtype,
+    ):
+        for filename in filenames:
+            state_dict = torch.load(filename, map_location="cpu")
+            for key, value in state_dict.items():
+                value = value.to(device if quantize is None else "cpu").to(dtype)
+
+                module_name, param_name = key.rsplit(".", 1)
+                module = model.get_submodule(module_name)
+
+                try:
+                    current_parameter_tensor = module._parameters[param_name]
+                    if current_parameter_tensor.shape != value.shape:
+                        raise ValueError(
+                            f"Name {key} -- Current {current_parameter_tensor.shape} and got {value.shape}"
+                        )
+                    module._parameters[param_name] = value
+                except KeyError:
+                    module._buffers[param_name] = value
+
+                del value
+
+        torch.cuda.empty_cache()
+        model.post_load_weights(quantize)
+
+#
+# class FlashNeoXSharded(FlashNeoX):
+#     def __init__(
+#         self,
+#         model_id: str,
+#         revision: Optional[str] = None,
+#         quantize: Optional[str] = None,
+#         trust_remote_code: bool = False,
+#     ):
+#         self.process_group, rank, world_size = initialize_torch_distributed()
+#         if torch.cuda.is_available():
+#             device = torch.device(f"cuda:{rank}")
+#             dtype = torch.float16
+#         else:
+#             raise NotImplementedError("FlashNeoX is only available on GPU")
+#
+#         tokenizer = AutoTokenizer.from_pretrained(
+#             model_id,
+#             revision=revision,
+#             padding_side="left",
+#             truncation_side="left",
+#             trust_remote_code=trust_remote_code,
+#         )
+#
+#         config = AutoConfig.from_pretrained(
+#             model_id, revision=revision, trust_remote_code=trust_remote_code
+#         )
+#
+#         torch.distributed.barrier(group=self.process_group)
+#         filenames = weight_files(model_id, revision=revision, extension=".safetensors")
+#
+#         with init_empty_weights():
+#             model = FlashGPTNeoXForCausalLM(config, self.process_group)
+#
+#         torch.distributed.barrier(group=self.process_group)
+#         self.load_weights(
+#             model,
+#             filenames,
+#             quantize=quantize,
+#             device=device,
+#             dtype=dtype,
+#             rank=rank,
+#             world_size=world_size,
+#         )
+#         torch.distributed.barrier(group=self.process_group)
+#         super(FlashCausalLM, self).__init__(
+#             model=model.to(device),
+#             tokenizer=tokenizer,
+#             requires_padding=False,
+#             dtype=dtype,
+#             device=device,
+#             rank=rank,
+#             world_size=world_size,
+#         )
+#
+#     @staticmethod
+#     def load_weights(
+#         model,
+#         filenames: List[str],
+#         quantize: Optional[str],
+#         device: torch.device,
+#         dtype: torch.dtype,
+#         rank: int,
+#         world_size: int,
+#     ):
+#         parameters = dict(model.named_parameters())
+#         for file in filenames:
+#             with safe_open(
+#                 file, framework="pt", device=str(device) if quantize is None else "cpu"
+#             ) as f:
+#                 for name in f.keys():
+#                     module_name, param_name = name.rsplit(".", 1)
+#                     module = model.get_submodule(module_name)
+#
+#                     current_parameter_tensor = parameters.get(name, None)
+#
+#                     slice_ = f.get_slice(name)
+#
+#                     if isinstance(module, TensorParallelColumnLinear):
+#                         size = slice_.get_shape()[0]
+#                         block_size = size // world_size
+#                         start = rank * block_size
+#                         stop = (rank + 1) * block_size
+#                         tensor = slice_[start:stop]
+#                     elif isinstance(module, TensorParallelRowLinear):
+#                         if param_name == "weight":
+#                             size = slice_.get_shape()[1]
+#                             block_size = size // world_size
+#                             start = rank * block_size
+#                             stop = (rank + 1) * block_size
+#                             tensor = slice_[:, start:stop]
+#                         else:
+#                             tensor = slice_[:]
+#                             # XXX: Hack for Rowlinear to add the bias only once.
+#                             if rank != 0:
+#                                 tensor = torch.zeros_like(tensor)
+#                     elif isinstance(module, TensorParallelEmbedding):
+#                         size = slice_.get_shape()[0]
+#                         block_size = size // world_size
+#                         start = rank * block_size
+#                         stop = (rank + 1) * block_size
+#                         tensor = slice_[start:stop]
+#                     elif name == "embed_out.weight" and model.gpt_neox.tp_embeddings:
+#                         size = slice_.get_shape()[0]
+#                         block_size = size // world_size
+#                         start = rank * block_size
+#                         stop = (rank + 1) * block_size
+#                         tensor = slice_[start:stop]
+#                     else:
+#                         try:
+#                             tensor = slice_[:]
+#                         except:
+#                             tensor = f.get_tensor(name)
+#
+#                     if (
+#                         current_parameter_tensor is not None
+#                         and current_parameter_tensor.shape != tensor.shape
+#                     ):
+#                         raise ValueError(
+#                             f"Name {name} -- Current {current_parameter_tensor.shape} and got {tensor.shape}"
+#                         )
+#
+#                     tensor = tensor.contiguous().to(dtype)
+#
+#                     if current_parameter_tensor is not None:
+#                         module._parameters[param_name] = tensor
+#                     else:
+#                         module._buffers[param_name] = tensor
+#
+#         model.post_load_weights(quantize)

server/text_generation_server/models/rw.py

@@ -0,0 +1,80 @@
+import torch
+
+from transformers import AutoTokenizer, AutoModelForCausalLM
+from typing import List, Optional, Tuple
+
+from text_generation_server.models import CausalLM
+
+
+class RW(CausalLM):
+    def __init__(
+            self,
+            model_id: str,
+            revision: Optional[str] = None,
+            quantize: Optional[str] = None,
+            trust_remote_code: bool = False,
+    ):
+        if torch.cuda.is_available():
+            device = torch.device("cuda")
+            dtype = torch.bfloat16
+        else:
+            if quantize:
+                raise ValueError("quantization is not available on CPU")
+
+            device = torch.device("cpu")
+            dtype = torch.float32
+
+        tokenizer = AutoTokenizer.from_pretrained(
+            model_id,
+            revision=revision,
+            padding_side="left",
+            truncation_side="left",
+            trust_remote_code=trust_remote_code,
+        )
+        model = AutoModelForCausalLM.from_pretrained(
+            model_id,
+            revision=revision,
+            torch_dtype=dtype,
+            device_map="auto"
+            if torch.cuda.is_available() and torch.cuda.device_count() > 1
+            else None,
+            load_in_8bit=quantize == "bitsandbytes",
+            trust_remote_code=trust_remote_code,
+        )
+        if torch.cuda.is_available() and torch.cuda.device_count() == 1:
+            model = model.cuda()
+
+        if tokenizer.pad_token_id is None:
+            if model.config.pad_token_id is not None:
+                tokenizer.pad_token_id = model.config.pad_token_id
+            elif model.config.eos_token_id is not None:
+                tokenizer.pad_token_id = model.config.eos_token_id
+            elif tokenizer.eos_token_id is not None:
+                tokenizer.pad_token_id = tokenizer.eos_token_id
+            else:
+                tokenizer.add_special_tokens({"pad_token": "[PAD]"})
+
+        super(CausalLM, self).__init__(
+            model=model,
+            tokenizer=tokenizer,
+            requires_padding=True,
+            dtype=dtype,
+            device=device,
+        )
+
+    def forward(
+            self, input_ids, attention_mask, position_ids, past_key_values: Optional = None
+    ) -> Tuple[torch.Tensor, List[Tuple[torch.Tensor, torch.Tensor]]]:
+        # Model Forward
+        if past_key_values is not None:
+            reshaped_past_key_values = []
+            for layer in past_key_values:
+                past_keys, past_values = layer
+                reshaped_past_key_values.append(
+                    (past_keys.view(-1, *past_keys.shape[-2:]), past_values.view(-1, *past_values.shape[-2:]))
+                )
+            past_key_values = reshaped_past_key_values
+
+        outputs = self.model.forward(input_ids=input_ids, attention_mask=attention_mask,
+                                     past_key_values=past_key_values)
+        return outputs.logits, outputs.past_key_values

server/text_generation_server/utils/layers.py

@@ -262,16 +262,13 @@ try:
             sin = torch.index_select(self._sin_cached, 0, position_ids)
             return cos.unsqueeze(1), sin.unsqueeze(1)
 
-        def forward(self, qkv: torch.Tensor, cos: torch.Tensor, sin: torch.Tensor):
+        def forward(self, x: torch.Tensor, cos: torch.Tensor, sin: torch.Tensor):
             rotary_dim = cos.shape[-1]
-            q1 = qkv[:, 0, :, :rotary_dim]
+            x1 = x[..., :rotary_dim]
-            q2 = qkv[:, 0, :, rotary_dim : 2 * rotary_dim]
+            x2 = x[..., rotary_dim : 2 * rotary_dim]
-            k1 = qkv[:, 1, :, :rotary_dim]
-            k2 = qkv[:, 1, :, rotary_dim : 2 * rotary_dim]
 
-            rotary_emb.apply_rotary(q1, q2, cos, sin, q1, q2, False)
+            rotary_emb.apply_rotary(x1, x2, cos, sin, x1, x2, False)
-            rotary_emb.apply_rotary(k1, k2, cos, sin, k1, k2, False)
+            return x
-            return qkv
 
 except ImportError:
     pass